title

Statistical Rethinking Winter 2019 Lecture 01

description

Lecture 01 of the Dec 2018 through March 2019 edition of Statistical Rethinking: A Bayesian Course with R and Stan.

detail

{'title': 'Statistical Rethinking Winter 2019 Lecture 01', 'heatmap': [{'end': 1049.017, 'start': 1005.863, 'weight': 0.713}, {'end': 1295.516, 'start': 1246.622, 'weight': 0.834}, {'end': 3230.29, 'start': 3147.761, 'weight': 0.736}, {'end': 3374.009, 'start': 3260.125, 'weight': 1}, {'end': 3480.571, 'start': 3438.668, 'weight': 0.761}], 'summary': 'Covers a 10-week course on bayesian statistics, practical bayesian inference, evolution and statistical testing, bayesian data analysis, bayesian inference, and probability and bayesian models, providing insights and analogies to emphasize the complexity of scientific phenomena, the need for skepticism in using statistical models, and the significance of bayesian data analysis in modern statistics.', 'chapters': [{'end': 399.87, 'segs': [{'end': 32.711, 'src': 'embed', 'start': 1.735, 'weight': 0, 'content': [{'end': 1.935, 'text': 'All right.', 'start': 1.735, 'duration': 0.2}, {'end': 2.575, 'text': 'Hi, everyone.', 'start': 2.095, 'duration': 0.48}, {'end': 3.256, 'text': "I'm Richard.", 'start': 2.875, 'duration': 0.381}, {'end': 6.457, 'text': 'For the next 10 weeks.', 'start': 4.456, 'duration': 2.001}, {'end': 11.319, 'text': 'well, not the next 10 weeks, but I will deliver you 10 weeks of instruction interrupted by the winter holidays.', 'start': 6.457, 'duration': 4.862}, {'end': 13.88, 'text': '10 weeks on Bayesian statistics.', 'start': 11.319, 'duration': 2.561}, {'end': 19.14, 'text': 'Before we talk about statistics, I want to talk about science.', 'start': 15.698, 'duration': 3.442}, {'end': 27.307, 'text': "I think, like me, most of you are here because you're scientists or researchers of another sort,", 'start': 20.121, 'duration': 7.186}, {'end': 32.711, 'text': 'and statistics is just some terrible thing you have to do to produce inferences.', 'start': 27.307, 'duration': 5.404}], 'summary': 'Richard will deliver 10 weeks of instruction on bayesian statistics, with interruptions for the winter holidays.', 'duration': 30.976, 'max_score': 1.735, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41735.jpg'}, {'end': 180.297, 'src': 'embed', 'start': 139.307, 'weight': 1, 'content': [{'end': 144.692, 'text': "whether you study the structure of the brain or, like me, you're interested in human societies and the evolution of organisms.", 'start': 139.307, 'duration': 5.385}, {'end': 151.766, 'text': 'All of these things are complicated processes that are not reducible really to individual statistical tests.', 'start': 145.444, 'duration': 6.322}, {'end': 158.368, 'text': 'And I think, then, the great disservice that the Society of Science, the Society of Research,', 'start': 153.106, 'duration': 5.262}, {'end': 168.931, 'text': 'more broadly has done is that most of the procedures we have canned for us on our computers are things that were invented to study well this agricultural trials in a particular part of England.', 'start': 158.368, 'duration': 10.563}, {'end': 176.976, 'text': "And this is a hard problem, but it's vastly simpler than the kinds of problems that most of us actually try to address.", 'start': 170.974, 'duration': 6.002}, {'end': 180.297, 'text': 'So things like t-tests and analysis of variance,', 'start': 177.276, 'duration': 3.021}], 'summary': 'Challenges in scientific research: complex processes defy simple statistical tests.', 'duration': 40.99, 'max_score': 139.307, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo4139307.jpg'}, {'end': 266.433, 'src': 'embed', 'start': 241.312, 'weight': 3, 'content': [{'end': 247.174, 'text': 'but you could search the internet for statistical decision tree and find any number of flowcharts similar to this.', 'start': 241.312, 'duration': 5.862}, {'end': 252.363, 'text': 'I want to say about these charts is that this is madness.', 'start': 250.262, 'duration': 2.101}, {'end': 253.965, 'text': 'This is pure madness.', 'start': 253.044, 'duration': 0.921}, {'end': 260.349, 'text': 'This is a great way to test undergraduates on something you can teach them, but this is no way to prepare people to do research.', 'start': 254.145, 'duration': 6.204}, {'end': 262.95, 'text': "There's no scientific principles in this.", 'start': 261.269, 'duration': 1.681}, {'end': 266.433, 'text': "It's just stuff about the data table you have.", 'start': 263.031, 'duration': 3.402}], 'summary': 'Statistical decision tree flowcharts lack scientific principles and are not suitable for research preparation.', 'duration': 25.121, 'max_score': 241.312, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo4241312.jpg'}, {'end': 312.131, 'src': 'embed', 'start': 284.802, 'weight': 5, 'content': [{'end': 288.066, 'text': 'something I know the biologists at least will have heard of Mann-Whitney U tests.', 'start': 284.802, 'duration': 3.264}, {'end': 289.527, 'text': 'Think of it as a little robot.', 'start': 288.466, 'duration': 1.061}, {'end': 291.189, 'text': 'It takes particular inputs.', 'start': 289.707, 'duration': 1.482}, {'end': 293.812, 'text': 'It does something with them and gives you an output.', 'start': 291.549, 'duration': 2.263}, {'end': 295.694, 'text': "It's like a Roomba, if you guys know a Roomba.", 'start': 293.932, 'duration': 1.762}, {'end': 297.456, 'text': 'It sweeps your floor.', 'start': 296.214, 'duration': 1.242}, {'end': 299.197, 'text': 'Little robots move around, sweep your floor.', 'start': 297.796, 'duration': 1.401}, {'end': 300.879, 'text': "Yeah? They're great, actually.", 'start': 299.337, 'duration': 1.542}, {'end': 306.845, 'text': 'You can reprogram them and do all kinds of things, like chase the cat and so on.', 'start': 300.919, 'duration': 5.926}, {'end': 312.131, 'text': 'These little robots, all of these tests are like little robots.', 'start': 309.488, 'duration': 2.643}], 'summary': 'Mann-whitney u tests are like little robots, taking inputs and producing outputs, similar to a roomba sweeping the floor.', 'duration': 27.329, 'max_score': 284.802, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo4284802.jpg'}, {'end': 384.521, 'src': 'embed', 'start': 359.18, 'weight': 4, 'content': [{'end': 366.062, 'text': "It's this thing about computers and robots as a special kind of computer that they're terrible at things that children are good at.", 'start': 359.18, 'duration': 6.882}, {'end': 370.003, 'text': "And they're really good at things that we're terrible at.", 'start': 368.083, 'duration': 1.92}, {'end': 371.765, 'text': 'Yeah, like playing Go.', 'start': 370.984, 'duration': 0.781}, {'end': 374.027, 'text': 'Computers are really good at playing the game of Go.', 'start': 372.325, 'duration': 1.702}, {'end': 376.59, 'text': 'Humans are bad at it.', 'start': 374.448, 'duration': 2.142}, {'end': 377.611, 'text': 'So we look at them in awe.', 'start': 376.63, 'duration': 0.981}, {'end': 380.474, 'text': 'Meanwhile, almost no robot can climb stairs.', 'start': 378.031, 'duration': 2.443}, {'end': 384.521, 'text': 'This turns out to be the cutting edge of artificial intelligence.', 'start': 382, 'duration': 2.521}], 'summary': 'Computers excel at games like go, while humans outperform in tasks like climbing stairs. this reveals the forefront of artificial intelligence.', 'duration': 25.341, 'max_score': 359.18, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo4359180.jpg'}], 'start': 1.735, 'title': 'Bayesian statistics and statistics misconceptions', 'summary': 'Introduces a 10-week course on bayesian statistics, emphasizing the complexity of scientific phenomena and the inadequacy of traditional statistical methods, while critiquing the use of decision trees in introductory stats classes and emphasizing the complementarity of computers and humans in data analysis and research.', 'chapters': [{'end': 219.068, 'start': 1.735, 'title': 'Bayesian statistics: making inferences in science', 'summary': 'Introduces a 10-week course on bayesian statistics, emphasizing the complexity of scientific phenomena and the inadequacy of traditional statistical methods for addressing these issues.', 'duration': 217.333, 'highlights': ['The complexity of the phenomena studied in science poses challenges that cannot be adequately addressed with traditional statistical tests The chapter emphasizes that the phenomena studied in science, such as the structure of the brain, human societies, and the evolution of organisms, are incredibly complicated and not reducible to individual statistical tests.', 'Introduction of a 10-week course on Bayesian statistics Richard introduces a 10-week course on Bayesian statistics, aimed at scientists and researchers to improve their ability to make inferences about the natural world.', 'Critique of traditional statistical methods and their inadequacy for addressing complex scientific problems The chapter highlights the limitations of traditional statistical methods, such as t-tests and analysis of variance, in addressing the complex problems encountered in scientific research.']}, {'end': 399.87, 'start': 221.164, 'title': 'Statistics misconceptions and robotic metaphor', 'summary': 'Critiques the use of decision trees in introductory stats classes, likening statistical tests to little robots and emphasizing the limitations and complementarity of computers and humans in data analysis and research.', 'duration': 178.706, 'highlights': ['The use of decision trees in introductory stats classes is critiqued for being insufficient to prepare people for research, lacking scientific principles, and being a disservice to understanding complex real-world phenomena like the border between Pakistan and India.', 'The speaker likens statistical tests to little robots, emphasizing their nature as tools with specific inputs and outputs, the potential for misbehavior if used incorrectly, and the complementarity between computers and humans in data analysis and research.', 'The limitations and capabilities of computers and robots are discussed, highlighting their strengths in areas where humans are weak, such as playing Go, but also pointing out their limitations, such as climbing stairs, which is considered the cutting edge of artificial intelligence research.']}], 'duration': 398.135, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41735.jpg', 'highlights': ['Introduction of a 10-week course on Bayesian statistics aimed at scientists and researchers', 'Critique of traditional statistical methods and their inadequacy for addressing complex scientific problems', 'The complexity of scientific phenomena poses challenges that cannot be adequately addressed with traditional statistical tests', 'The use of decision trees in introductory stats classes is critiqued for being insufficient to prepare people for research', 'The limitations and capabilities of computers and robots are discussed, highlighting their strengths in areas where humans are weak', 'The speaker likens statistical tests to little robots, emphasizing their nature as tools with specific inputs and outputs']}, {'end': 1057.239, 'segs': [{'end': 555.848, 'src': 'embed', 'start': 519.464, 'weight': 4, 'content': [{'end': 520.225, 'text': 'And this is a lesson.', 'start': 519.464, 'duration': 0.761}, {'end': 525.648, 'text': "It's a moral lesson about, in the Hebraic tradition, of taking God's power of creation into your own hands.", 'start': 520.245, 'duration': 5.403}, {'end': 528.37, 'text': "Even if it's for good purposes, it's just too dangerous.", 'start': 526.208, 'duration': 2.162}, {'end': 530.471, 'text': 'Mortals cannot have this power of creation.', 'start': 528.87, 'duration': 1.601}, {'end': 535.014, 'text': "In this class, we're going to use the power of creation, and we're going to make statistical models.", 'start': 531.011, 'duration': 4.003}, {'end': 538.095, 'text': "But I want you to remember the golem, because you're making little golems.", 'start': 535.314, 'duration': 2.781}, {'end': 542.678, 'text': 'And their havoc will be isolated to your laptop, for the most part, I think.', 'start': 538.756, 'duration': 3.922}, {'end': 545.1, 'text': 'But nevertheless, you will feel their destructive power.', 'start': 543.359, 'duration': 1.741}, {'end': 555.848, 'text': 'And what I hope is, by the end of the course, you have enough skill and wisdom that you can use them responsibly and usefully,', 'start': 546.921, 'duration': 8.927}], 'summary': "Lesson on not wielding god's power; creating statistical models like golems.", 'duration': 36.384, 'max_score': 519.464, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo4519464.jpg'}, {'end': 720.699, 'src': 'embed', 'start': 676.601, 'weight': 0, 'content': [{'end': 680.242, 'text': 'So statistical models are not true or false things.', 'start': 676.601, 'duration': 3.641}, {'end': 683.482, 'text': "They're not rational entities.", 'start': 680.262, 'duration': 3.22}, {'end': 689.224, 'text': 'Their behavior does have a logic and does follow a logic, and that lets us design them and analyze why they behave the way they do.', 'start': 683.742, 'duration': 5.482}, {'end': 690.524, 'text': "But they're just tools.", 'start': 689.684, 'duration': 0.84}, {'end': 694.945, 'text': 'So they can neither be false or true any more than a hammer can be false or true.', 'start': 691.084, 'duration': 3.861}, {'end': 699.906, 'text': "There may be better tools for making a table than the hammer you have at hand, but the hammer isn't false.", 'start': 695.465, 'duration': 4.441}, {'end': 703.188, 'text': "It's just worse than the ideal hammer you'd like to have.", 'start': 700.606, 'duration': 2.582}, {'end': 710.372, 'text': 'So let me talk a little bit about what you should expect from this course.', 'start': 705.709, 'duration': 4.663}, {'end': 713.715, 'text': 'I have prepared 10 weeks of material.', 'start': 712.214, 'duration': 1.501}, {'end': 715.196, 'text': 'This is 20 lectures.', 'start': 714.115, 'duration': 1.081}, {'end': 716.116, 'text': 'This is two hours a week.', 'start': 715.256, 'duration': 0.86}, {'end': 720.699, 'text': "There's also a book that goes with it and some software.", 'start': 716.596, 'duration': 4.103}], 'summary': 'Statistical models are logical tools, not true or false, with 10 weeks of material, 20 lectures, and two hours per week.', 'duration': 44.098, 'max_score': 676.601, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo4676601.jpg'}, {'end': 920.944, 'src': 'embed', 'start': 896.77, 'weight': 1, 'content': [{'end': 902.654, 'text': 'None of the problems that are relevant to any of you or to me are solvable analytically with analytical mathematics.', 'start': 896.77, 'duration': 5.884}, {'end': 904.174, 'text': 'We have to use code to do it.', 'start': 902.794, 'duration': 1.38}, {'end': 905.815, 'text': 'So those are the skills we emphasize.', 'start': 904.435, 'duration': 1.38}, {'end': 913.58, 'text': 'And the exercises are structured so that I hope you get enough confidence to be comfortable being confused for the rest of your life.', 'start': 906.336, 'duration': 7.244}, {'end': 917.903, 'text': "Because that's what it's like when you're working on hard problems.", 'start': 914.821, 'duration': 3.082}, {'end': 920.944, 'text': 'Nobody knows the answer to is you feel constantly confused.', 'start': 918.283, 'duration': 2.661}], 'summary': 'Problem-solving skills emphasized through coding, fostering comfort with confusion in difficult problems.', 'duration': 24.174, 'max_score': 896.77, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo4896770.jpg'}, {'end': 959.539, 'src': 'embed', 'start': 935.239, 'weight': 2, 'content': [{'end': 942.181, 'text': "Again, whether you're a scientist like me, working for the public, or whether you're a private researcher, it's the same game.", 'start': 935.239, 'duration': 6.942}, {'end': 943.902, 'text': "There's a lot of confusion.", 'start': 942.822, 'duration': 1.08}, {'end': 948.904, 'text': "If you're studying a problem no one knows the answer to, you should expect to be confused and frustrated on most days of the week.", 'start': 943.942, 'duration': 4.962}, {'end': 950.357, 'text': "That's perfectly normal.", 'start': 949.497, 'duration': 0.86}, {'end': 952.678, 'text': 'And so I want you to get used to that in this class.', 'start': 950.897, 'duration': 1.781}, {'end': 959.539, 'text': 'And I will give you homework assignments that make you feel a little bit confused at first when you start, but then they are solvable.', 'start': 953.838, 'duration': 5.701}], 'summary': 'Scientist emphasizes the normalcy of confusion in research and provides solvable homework assignments.', 'duration': 24.3, 'max_score': 935.239, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo4935239.jpg'}, {'end': 1049.017, 'src': 'heatmap', 'start': 1005.863, 'weight': 0.713, 'content': [{'end': 1010.386, 'text': "And you'll be working with the experimental branch, which I know sounds very exciting right?", 'start': 1005.863, 'duration': 4.523}, {'end': 1018.052, 'text': "Because there are new features in here that I'm building into the second edition of the book that I think are very good.", 'start': 1011.247, 'duration': 6.805}, {'end': 1022.195, 'text': "They're cool features, and I want to try them out on all of you and see how you use them.", 'start': 1018.092, 'duration': 4.103}, {'end': 1028.459, 'text': "And also with this, there's the second edition of the book, which is up on my website.", 'start': 1023.456, 'duration': 5.003}, {'end': 1029.8, 'text': 'You need a password.', 'start': 1029, 'duration': 0.8}, {'end': 1033.403, 'text': 'The password is my favorite 1980s TV character, Blossom.', 'start': 1029.819, 'duration': 3.584}, {'end': 1036.654, 'text': "There's one person who knows this joke.", 'start': 1035.054, 'duration': 1.6}, {'end': 1049.017, 'text': "OK So what's going on with the second edition? Some of you are taking this course for the second time.", 'start': 1042.076, 'duration': 6.941}], 'summary': "Experimental branch has new features for second edition of the book. second edition of the book is available on website with password 'blossom'.", 'duration': 43.154, 'max_score': 1005.863, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41005863.jpg'}], 'start': 399.87, 'title': 'Statistical models and practical bayesian inference', 'summary': 'Discusses the analogy of statistical models with the golem myth, emphasizing the need for skepticism and skill in their use, while also detailing a 10-week course on practical bayesian inference consisting of 20 lectures, 2 hours per week, fostering confidence in solving complex problems.', 'chapters': [{'end': 694.945, 'start': 399.87, 'title': 'The golem myth: lessons for statistical models', 'summary': 'Discusses the analogy of statistical models with the golem myth, emphasizing the responsibility of using them wisely due to their powerful but blind nature, and the need for skepticism and skill in their use.', 'duration': 295.075, 'highlights': ['The Golem analogy emphasizes the need for responsibility and circumspection in using statistical models due to their powerful but blind nature. Emphasizes the responsibility of using statistical models wisely due to their powerful but blind nature.', "Statistical models, like the Golem, are blind to the creator's intent and take instructions literally, requiring careful use due to their potential for misuse. Emphasizes the blind nature of statistical models and the potential for misuse due to literal interpretation of instructions.", 'The chapter compares statistical models to the Golem, emphasizing that models are not rational entities, but rather tools with behavior that can be analyzed for usefulness. Compares statistical models to the Golem and emphasizes that models are tools with behavior that can be analyzed for usefulness.']}, {'end': 1057.239, 'start': 695.465, 'title': 'Practical bayesian inference', 'summary': 'Entails a 10-week course consisting of 20 lectures, 2 hours per week, aiming to deliver practical model building and criticizing skills through coding exercises, fostering confidence to work through confusion in solving complex problems.', 'duration': 361.774, 'highlights': ['The course consists of 10 weeks with 20 lectures, 2 hours per week, and is accompanied by a book and software. The course provides a structured 10-week curriculum with 20 lectures and 2 hours of teaching per week, along with supplementary materials such as a book and software.', 'Emphasis is placed on practical model building and criticizing skills, predominantly through coding exercises. The focus of the course is on delivering practical model building and criticizing skills, primarily through coding exercises rather than analytical mathematics.', 'The course aims to instill confidence in working through confusion when solving complex problems, essential for researchers. The course aims to build confidence in dealing with confusion when tackling complex problems, a crucial skill for researchers.']}], 'duration': 657.369, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo4399870.jpg', 'highlights': ['The course provides a structured 10-week curriculum with 20 lectures and 2 hours of teaching per week, along with supplementary materials such as a book and software.', 'The focus of the course is on delivering practical model building and criticizing skills, primarily through coding exercises rather than analytical mathematics.', 'The course aims to build confidence in dealing with confusion when tackling complex problems, a crucial skill for researchers.', 'Compares statistical models to the Golem and emphasizes that models are tools with behavior that can be analyzed for usefulness.', 'Emphasizes the blind nature of statistical models and the potential for misuse due to literal interpretation of instructions.', 'Emphasizes the responsibility of using statistical models wisely due to their powerful but blind nature.']}, {'end': 1741.062, 'segs': [{'end': 1093.884, 'src': 'embed', 'start': 1057.339, 'weight': 0, 'content': [{'end': 1067.607, 'text': 'So things keep evolving here and I think The problem with publishing a book is it kind of freezes an evolutionary process at a particular point in time.', 'start': 1057.339, 'duration': 10.268}, {'end': 1070.65, 'text': 'But every time I teach the course, I figure out how to do something better, I think.', 'start': 1067.687, 'duration': 2.963}, {'end': 1073.313, 'text': 'And so second edition is helping me do that.', 'start': 1070.99, 'duration': 2.323}, {'end': 1076.216, 'text': 'So you get the second edition for free as a PDF.', 'start': 1073.573, 'duration': 2.643}, {'end': 1079.739, 'text': "And what do you get for this? Well, you're going to get a large number of typos.", 'start': 1076.576, 'duration': 3.163}, {'end': 1087.362, 'text': "But in addition to that, You're going to get some really useful stuff, like prior predictive simulation, and you won't know what that means right now,", 'start': 1079.819, 'duration': 7.543}, {'end': 1093.884, 'text': 'but it will help you to understand priors way better than previous versions of the book would have allowed you to do so.', 'start': 1087.362, 'duration': 6.522}], 'summary': 'Second edition of the book offers free pdf with improvements, including prior predictive simulation.', 'duration': 36.545, 'max_score': 1057.339, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41057339.jpg'}, {'end': 1295.516, 'src': 'heatmap', 'start': 1246.622, 'weight': 0.834, 'content': [{'end': 1257.178, 'text': 'So the thing about them that I want to spend a few slides trying to talk you out of is the idea that the purpose of a statistical procedure is to falsify a hypothesis.', 'start': 1246.622, 'duration': 10.556}, {'end': 1265.344, 'text': "I think it's fine to say that the purpose of a research paradigm is to falsify a hypothesis.", 'start': 1258.902, 'duration': 6.442}, {'end': 1265.844, 'text': "That's okay.", 'start': 1265.384, 'duration': 0.46}, {'end': 1268.845, 'text': 'But the purpose of a statistical test is definitely not.', 'start': 1266.384, 'duration': 2.461}, {'end': 1272.246, 'text': "Or at least, if that's the purpose you put it to, you're going to be disappointed.", 'start': 1269.165, 'duration': 3.081}, {'end': 1273.867, 'text': "So here's why.", 'start': 1273.226, 'duration': 0.641}, {'end': 1278.848, 'text': 'Let me give you an example from my own area of research, or say my own broader field.', 'start': 1274.247, 'duration': 4.601}, {'end': 1281.099, 'text': 'But I imagine you can think of your own.', 'start': 1279.677, 'duration': 1.422}, {'end': 1286.045, 'text': 'The basic problem is that statistical models are not hypotheses and they do not embody them.', 'start': 1281.459, 'duration': 4.586}, {'end': 1288.207, 'text': "It's much more complicated than that.", 'start': 1286.926, 'duration': 1.281}, {'end': 1295.516, 'text': 'So it used to be, last century, there was a big fight in population genetics over whether evolution was neutral.', 'start': 1288.748, 'duration': 6.768}], 'summary': 'Statistical procedures do not aim to falsify hypotheses; they are not meant for that purpose.', 'duration': 48.894, 'max_score': 1246.622, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41246622.jpg'}, {'end': 1299.618, 'src': 'embed', 'start': 1269.165, 'weight': 1, 'content': [{'end': 1272.246, 'text': "Or at least, if that's the purpose you put it to, you're going to be disappointed.", 'start': 1269.165, 'duration': 3.081}, {'end': 1273.867, 'text': "So here's why.", 'start': 1273.226, 'duration': 0.641}, {'end': 1278.848, 'text': 'Let me give you an example from my own area of research, or say my own broader field.', 'start': 1274.247, 'duration': 4.601}, {'end': 1281.099, 'text': 'But I imagine you can think of your own.', 'start': 1279.677, 'duration': 1.422}, {'end': 1286.045, 'text': 'The basic problem is that statistical models are not hypotheses and they do not embody them.', 'start': 1281.459, 'duration': 4.586}, {'end': 1288.207, 'text': "It's much more complicated than that.", 'start': 1286.926, 'duration': 1.281}, {'end': 1295.516, 'text': 'So it used to be, last century, there was a big fight in population genetics over whether evolution was neutral.', 'start': 1288.748, 'duration': 6.768}, {'end': 1299.618, 'text': 'And what that would mean is, of course, everyone believed that natural selection mattered,', 'start': 1296.454, 'duration': 3.164}], 'summary': 'Statistical models are not hypotheses; a big fight in population genetics over whether evolution was neutral occurred last century.', 'duration': 30.453, 'max_score': 1269.165, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41269165.jpg'}, {'end': 1458.203, 'src': 'embed', 'start': 1425.07, 'weight': 3, 'content': [{'end': 1426.431, 'text': 'It could matter in this way or that way.', 'start': 1425.07, 'duration': 1.361}, {'end': 1429.233, 'text': 'It could affect chromosomes, all kinds of stuff it could do.', 'start': 1426.451, 'duration': 2.782}, {'end': 1431.394, 'text': 'And so you have to make multiple process models.', 'start': 1429.693, 'duration': 1.701}, {'end': 1433.116, 'text': 'Well, it turned out early on.', 'start': 1431.915, 'duration': 1.201}, {'end': 1440.381, 'text': "a fellow named John Gillespie showed that there's a particular version of selection matters, the fluctuating selection model at the very bottom,", 'start': 1433.116, 'duration': 7.265}, {'end': 1444.324, 'text': 'which produces exactly the same statistical predictions as the neutral model.', 'start': 1440.381, 'duration': 3.943}, {'end': 1453.179, 'text': 'This is not an unusual situation in the sciences for anything more complicated than a trial of one barley variety versus another.', 'start': 1445.912, 'duration': 7.267}, {'end': 1458.203, 'text': 'Is that multiple process models make exactly the same statistical predictions.', 'start': 1454.72, 'duration': 3.483}], 'summary': 'Multiple process models produce same statistical predictions as neutral model.', 'duration': 33.133, 'max_score': 1425.07, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41425070.jpg'}, {'end': 1647.158, 'src': 'embed', 'start': 1620.922, 'weight': 4, 'content': [{'end': 1627.186, 'text': "There's lots of things about evidence that requires having more than one model and seeing which is consistent with the predictions you see.", 'start': 1620.922, 'duration': 6.264}, {'end': 1629.647, 'text': "It's not enough to just falsify things.", 'start': 1627.846, 'duration': 1.801}, {'end': 1634.65, 'text': 'You have to build a substantive theory at some point, not just knock down null models.', 'start': 1629.767, 'duration': 4.883}, {'end': 1643.376, 'text': "That aside, in Popper's view of falsification, you're falsifying the explanatory model, not some null model of zero effect.", 'start': 1636.571, 'duration': 6.805}, {'end': 1647.158, 'text': 'In statistical procedures, like in that diagram I showed you, this has all been reversed.', 'start': 1643.456, 'duration': 3.702}], 'summary': "Evidence requires multiple models to test consistency. falsification is not enough; substantive theories must be built. popper's view is different from statistical procedures.", 'duration': 26.236, 'max_score': 1620.922, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41620922.jpg'}, {'end': 1729.213, 'src': 'embed', 'start': 1702.939, 'weight': 5, 'content': [{'end': 1706.941, 'text': "Let's get a little bit more practical, give you some expectations about what we're going to learn.", 'start': 1702.939, 'duration': 4.002}, {'end': 1713.606, 'text': 'So as I said, I want you to become Golem engineers, and entry level Golem engineers.', 'start': 1708.422, 'duration': 5.184}, {'end': 1718.786, 'text': 'We need some framework and some set of principles that we can build our own statistical models.', 'start': 1714.884, 'duration': 3.902}, {'end': 1725.09, 'text': "We're going to go into this not with the idea that we're going to be selecting from some toolbox of pre-made robots.", 'start': 1719.207, 'duration': 5.883}, {'end': 1729.213, 'text': "We're going to build our own, and you're going to learn the principles by which they're constructed,", 'start': 1725.41, 'duration': 3.803}], 'summary': 'Learn to become entry-level golem engineers and build statistical models.', 'duration': 26.274, 'max_score': 1702.939, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41702939.jpg'}], 'start': 1057.339, 'title': 'Evolution and statistical testing', 'summary': 'Delves into the evolution of statistical models, highlighting new features in the second edition and emphasizing the need to understand statistical models as tools. it also discusses the historical debate between neutral and selection matters models in population genetics, emphasizing the need for multiple process and statistical models to make predictions and distinguish between them.', 'chapters': [{'end': 1288.207, 'start': 1057.339, 'title': 'Evolution of statistical models', 'summary': 'Discusses the evolution of statistical models, introducing the second edition of a book with new features like prior predictive simulation and causal inference, emphasizing the need to understand statistical models as tools rather than rational entities and highlighting the limitations of statistical tests in falsifying hypotheses.', 'duration': 230.868, 'highlights': ['The second edition of the book offers new features like prior predictive simulation and causal inference, providing a better understanding of statistical models. The second edition of the book introduces new features such as prior predictive simulation and causal inference, offering a better understanding of statistical models for the readers.', 'Emphasizes the limitations of statistical tests in falsifying hypotheses and underscores the need to view statistical models as tools rather than rational entities. The chapter emphasizes the limitations of statistical tests in falsifying hypotheses and highlights the need to view statistical models as tools rather than rational entities.', 'Discusses the evolution of statistical models and the continuous improvement process, which challenges the idea of freezing the evolutionary process through book publication. The chapter discusses the continuous improvement process in statistical models, challenging the idea of freezing the evolutionary process through book publication.']}, {'end': 1741.062, 'start': 1288.748, 'title': 'Evolutionary models and statistical testing', 'summary': 'Discusses the historical debate between neutral and selection matters models in population genetics, emphasizing the need for multiple process and statistical models to make predictions and distinguish between them, as well as the importance of building substantive research hypotheses with point predictions and trying to falsify those, rather than falsifying the null model of zero effect.', 'duration': 452.314, 'highlights': ['The historical debate between neutral and selection matters models in population genetics emphasizes the need for multiple process and statistical models to distinguish between them. The chapter discusses the historical debate between neutral and selection matters models in population genetics, emphasizing the need for multiple process and statistical models to make predictions and distinguish between them.', 'The importance of building substantive research hypotheses with point predictions and trying to falsify those, rather than falsifying the null model of zero effect. The chapter emphasizes the importance of building substantive research hypotheses with point predictions and trying to falsify those, rather than falsifying the null model of zero effect, as well as the need for confirmation and building a substantive theory.', 'The need for Golem engineers to build their own statistical models and learn the principles to criticize and refine them. The chapter highlights the need for Golem engineers to build their own statistical models and learn the principles to criticize and refine them, rather than selecting from a pre-made toolbox of models.']}], 'duration': 683.723, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41057339.jpg', 'highlights': ['The second edition of the book introduces new features such as prior predictive simulation and causal inference, offering a better understanding of statistical models for the readers.', 'The chapter emphasizes the limitations of statistical tests in falsifying hypotheses and highlights the need to view statistical models as tools rather than rational entities.', 'The chapter discusses the continuous improvement process in statistical models, challenging the idea of freezing the evolutionary process through book publication.', 'The chapter discusses the historical debate between neutral and selection matters models in population genetics, emphasizing the need for multiple process and statistical models to make predictions and distinguish between them.', 'The chapter emphasizes the importance of building substantive research hypotheses with point predictions and trying to falsify those, rather than falsifying the null model of zero effect, as well as the need for confirmation and building a substantive theory.', 'The chapter highlights the need for Golem engineers to build their own statistical models and learn the principles to criticize and refine them, rather than selecting from a pre-made toolbox of models.']}, {'end': 2254.431, 'segs': [{'end': 1840.555, 'src': 'embed', 'start': 1814.252, 'weight': 0, 'content': [{'end': 1818.882, 'text': 'And so really we waited until the 1980s when we could put Markov chains on the desktop.', 'start': 1814.252, 'duration': 4.63}, {'end': 1824.145, 'text': 'And this has led to a revolution in the use of Bayesian methods since then.', 'start': 1819.182, 'duration': 4.963}, {'end': 1825.906, 'text': 'There was this project called the BUGS project.', 'start': 1824.185, 'duration': 1.721}, {'end': 1827.767, 'text': 'Some of you may have heard of it.', 'start': 1826.366, 'duration': 1.401}, {'end': 1830.749, 'text': 'Bayesian inference using Gibbs sampling is what BUGS stands for.', 'start': 1828.227, 'duration': 2.522}, {'end': 1832.71, 'text': 'BUGS is now kind of deceased.', 'start': 1831.269, 'duration': 1.441}, {'end': 1833.851, 'text': "It's old rickety software.", 'start': 1832.79, 'duration': 1.061}, {'end': 1840.555, 'text': "It's been replaced by newer things, but it was a heroic project and it led to a revolution,", 'start': 1833.891, 'duration': 6.664}], 'summary': 'In the 1980s, markov chains on desktops revolutionized bayesian methods, exemplified by the bugs project.', 'duration': 26.303, 'max_score': 1814.252, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41814252.jpg'}, {'end': 1881.495, 'src': 'embed', 'start': 1853.526, 'weight': 1, 'content': [{'end': 1860.489, 'text': "And in this course, I'm going to ignore this controversy except for this because in stats departments, this controversy is basically over.", 'start': 1853.526, 'duration': 6.963}, {'end': 1865.251, 'text': 'Stats departments the world over are essentially Bayesian now.', 'start': 1861.49, 'duration': 3.761}, {'end': 1867.432, 'text': 'Most of their research is Bayesian.', 'start': 1865.431, 'duration': 2.001}, {'end': 1870.974, 'text': 'Stats taught to undergrads is mainly non-Bayesian.', 'start': 1868.593, 'duration': 2.381}, {'end': 1872.855, 'text': 'And this is a weird friction.', 'start': 1871.714, 'duration': 1.141}, {'end': 1876.272, 'text': 'And in time this might change.', 'start': 1873.97, 'duration': 2.302}, {'end': 1881.495, 'text': "And it's still, there are still scientific fields in which it's controversial to be Bayesian.", 'start': 1877.633, 'duration': 3.862}], 'summary': 'Stats departments worldwide are predominantly bayesian, with most research being bayesian, but there is still friction as undergrad stats are mainly non-bayesian.', 'duration': 27.969, 'max_score': 1853.526, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41853526.jpg'}, {'end': 2028.832, 'src': 'embed', 'start': 2004.812, 'weight': 2, 'content': [{'end': 2012.555, 'text': 'What is the causal process going on? Then you see some observations which are presumably consequences of that process.', 'start': 2004.812, 'duration': 7.743}, {'end': 2016.296, 'text': "And you then say, OK, I've got alternative sets of assumptions.", 'start': 2013.215, 'duration': 3.081}, {'end': 2018.577, 'text': 'One of these sets of assumptions.', 'start': 2017.416, 'duration': 1.161}, {'end': 2025.611, 'text': "It is much more plausible to see this than the others, and that's all Bayesian data analysis does, but it does it in a very specific, counting way.", 'start': 2019.828, 'duration': 5.783}, {'end': 2028.832, 'text': 'You have to actually count stuff up, or rather your golem does.', 'start': 2026.231, 'duration': 2.601}], 'summary': 'Bayesian data analysis counts and compares alternative assumptions to determine plausibility.', 'duration': 24.02, 'max_score': 2004.812, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo42004812.jpg'}, {'end': 2083.03, 'src': 'embed', 'start': 2046.331, 'weight': 3, 'content': [{'end': 2051.132, 'text': "Multi-level models, we'll wait until the second half of the course, multi-level models are models in models.", 'start': 2046.331, 'duration': 4.801}, {'end': 2054.793, 'text': 'So if you liked models, well good news, we put models in your models.', 'start': 2052.172, 'duration': 2.621}, {'end': 2063.916, 'text': "And what's nice about this is it lets you deal in a very transparent way with lots of really routine scientific problems like measurement error,", 'start': 2055.893, 'duration': 8.023}, {'end': 2066.356, 'text': 'missing data, heterogeneity of effects.', 'start': 2063.916, 'duration': 2.44}, {'end': 2069.297, 'text': 'All of those come from embedding models within models.', 'start': 2066.916, 'duration': 2.381}, {'end': 2083.03, 'text': 'So as a set of examples, you want to deal with repeat sampling or imbalance sampling, variation across studies or between subjects within studies.', 'start': 2072.697, 'duration': 10.333}], 'summary': 'Multi-level models embed models within models to handle measurement error, missing data, and heterogeneity of effects.', 'duration': 36.699, 'max_score': 2046.331, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo42046331.jpg'}], 'start': 1741.823, 'title': 'Bayesian data analysis', 'summary': 'Provides an overview of bayesian data analysis, emphasizing its historical significance, computational challenges, widespread adoption in modern statistics, and controversies surrounding its use. it also explains the core concept of bayesian data analysis, the significance of multilevel models, model comparison, and guarding against overfitting.', 'chapters': [{'end': 1939.165, 'start': 1741.823, 'title': 'Bayesian data analysis overview', 'summary': 'Provides an overview of bayesian data analysis, emphasizing its historical significance, computational challenges, and widespread adoption in modern statistics, with a brief mention of its developers and the controversies surrounding its use.', 'duration': 197.342, 'highlights': ['Bayesian data analysis is the original statistics before the imperialism of frequentist English statisticians and Ronald Fisher, and its underdevelopment for centuries was largely due to the lack of good computers, until the 1980s revolutionized its use with the introduction of Markov chains and the BUGS project.', 'The BUGS project, which stands for Bayesian inference using Gibbs sampling, was a heroic project that revolutionized the use of Bayesian methods in stats departments, applied stats programs, and private research, leading to Bayesian methods being widely adopted in stats departments worldwide.', 'Stats departments worldwide are essentially Bayesian now, with most of their research being Bayesian, while undergraduate statistics education is mainly non-Bayesian, creating a friction between the two approaches, which is considered a historical issue that is not necessary to dwell on in the present day.', 'The developers of Bayesian inference, including Laplace, Harold Jeffries, and Bertha Swirls, made significant contributions to the development and propagation of the Bayesian approach, with Laplace being the most influential figure in its development.']}, {'end': 2254.431, 'start': 1939.165, 'title': 'Bayesian data analysis & multilevel models', 'summary': 'Explains the core concept of bayesian data analysis, which involves counting ways data can happen according to assumptions and the significance of multilevel models for dealing with scientific problems, along with the importance of model comparison and guarding against overfitting.', 'duration': 315.266, 'highlights': ['Bayesian data analysis involves counting ways data can happen according to assumptions, making assumptions about the causal process, and then comparing these assumptions to determine plausibility. None', 'Multilevel models are useful for dealing with routine scientific problems like measurement error, missing data, and heterogeneity of effects, and they involve embedding models within models. None', 'Model comparison is essential for comparing multiple explanatory hypotheses and guarding against overfitting, which is a significant phenomenon in statistics. None']}], 'duration': 512.608, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo41741823.jpg', 'highlights': ['The BUGS project revolutionized the use of Bayesian methods in stats departments and private research.', 'Stats departments worldwide are essentially Bayesian now, with most of their research being Bayesian.', 'Bayesian data analysis involves counting ways data can happen according to assumptions and comparing these assumptions to determine plausibility.', 'Multilevel models are useful for dealing with routine scientific problems like measurement error, missing data, and heterogeneity of effects.']}, {'end': 3146.02, 'segs': [{'end': 2340.66, 'src': 'embed', 'start': 2307.249, 'weight': 1, 'content': [{'end': 2311.304, 'text': 'which makes the Earth much smaller than it actually is.', 'start': 2307.249, 'duration': 4.055}, {'end': 2316.631, 'text': 'So this is ironic, of course, because the ancient Greeks knew how big the Earth was right?', 'start': 2311.725, 'duration': 4.906}, {'end': 2321.272, 'text': 'They used shadows and wells to figure out how the circumference of the Earth very accurately.', 'start': 2316.851, 'duration': 4.421}, {'end': 2331.116, 'text': 'But Colombo obeyed an Austrian geometer, Boeheim, who we have someone in here with the same last name, probably a descendant.', 'start': 2321.813, 'duration': 9.303}, {'end': 2331.517, 'text': "I don't know.", 'start': 2331.176, 'duration': 0.341}, {'end': 2340.66, 'text': 'And who, for his own reasons that are lost in time, decided to make the Earth smaller.', 'start': 2333.317, 'duration': 7.343}], 'summary': "Ancient greeks accurately measured earth's circumference using shadows and wells, but columbus mistakenly made it smaller.", 'duration': 33.411, 'max_score': 2307.249, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo42307249.jpg'}, {'end': 2452.873, 'src': 'embed', 'start': 2423.395, 'weight': 0, 'content': [{'end': 2424.816, 'text': 'And it interfered with his predictions.', 'start': 2423.395, 'duration': 1.421}, {'end': 2429.078, 'text': 'And he ended up in a very different place than he expected to.', 'start': 2425.236, 'duration': 3.842}, {'end': 2433.221, 'text': 'He was lucky he lived, right? I mean, the expected result is he just would have died at sea like most sailors did.', 'start': 2429.318, 'duration': 3.903}, {'end': 2441.824, 'text': 'So I want you to consider this bizarre analogy and keep it in your mind, to think about when you make a model.', 'start': 2435.239, 'duration': 6.585}, {'end': 2444.987, 'text': "you're like Christopher Columbus planning with this Austrian globe.", 'start': 2441.824, 'duration': 3.163}, {'end': 2447.489, 'text': "And you're betting your life on it.", 'start': 2445.788, 'duration': 1.701}, {'end': 2449.591, 'text': "Well, you won't be betting your life on it.", 'start': 2448.41, 'duration': 1.181}, {'end': 2452.873, 'text': "But you're betting some prediction on it.", 'start': 2451.432, 'duration': 1.441}], 'summary': 'Modeling is like columbus planning with a globe; betting some prediction on it.', 'duration': 29.478, 'max_score': 2423.395, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo42423395.jpg'}, {'end': 2495.862, 'src': 'embed', 'start': 2464.032, 'weight': 3, 'content': [{'end': 2472.037, 'text': "and we're going to learn ways to study the output of our models so that we can discover these effects and reconsider them.", 'start': 2464.032, 'duration': 8.005}, {'end': 2477.321, 'text': 'This is what is called in Bayesian inference, the small world-large world distinction,', 'start': 2473.018, 'duration': 4.303}, {'end': 2485.567, 'text': 'and this is from an influential Bayesian statistician from the middle of last century, Leonard Jimmy Savage, L.J.', 'start': 2477.321, 'duration': 8.246}, {'end': 2495.862, 'text': 'Savage had a book in 1954 which was very influential in the development of Bayesian inference in the second half of the 20th century,', 'start': 2485.587, 'duration': 10.275}], 'summary': "Studying model output is crucial in bayesian inference, influenced by l.j. savage's 1954 book.", 'duration': 31.83, 'max_score': 2464.032, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo42464032.jpg'}, {'end': 2940.883, 'src': 'embed', 'start': 2917.45, 'weight': 4, 'content': [{'end': 2925.273, 'text': 'And then finally, the case where there are three blue marbles and one white marbles, there are nine ways to make it happen.', 'start': 2917.45, 'duration': 7.823}, {'end': 2928.974, 'text': "So now you can compare, and you've got these relative counts.", 'start': 2926.192, 'duration': 2.782}, {'end': 2931.736, 'text': 'And this is Bayesian inference, and that is it.', 'start': 2929.955, 'duration': 1.781}, {'end': 2938.121, 'text': "And so every time someone says they've done a Bayesian analysis, I want you to think about the bag of marbles, because that's a Bayesian model,", 'start': 2932.036, 'duration': 6.085}, {'end': 2940.883, 'text': 'no matter how complicated just counting marbles.', 'start': 2938.121, 'duration': 2.762}], 'summary': 'Bayesian inference: 3 blue, 1 white marbles, 9 ways to happen', 'duration': 23.433, 'max_score': 2917.45, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo42917450.jpg'}, {'end': 3048.305, 'src': 'embed', 'start': 3021.865, 'weight': 5, 'content': [{'end': 3026.508, 'text': 'But the wonderful thing about Bayesian inference is you can just take your previous counts and update them with the new count.', 'start': 3021.865, 'duration': 4.643}, {'end': 3027.728, 'text': 'This is called Bayesian updating.', 'start': 3026.548, 'duration': 1.18}, {'end': 3033.352, 'text': "There's nothing more than taking the previous counts and multiplying by the new counts.", 'start': 3028.329, 'duration': 5.023}, {'end': 3037.114, 'text': "And I'll show you why it's multiplication later.", 'start': 3034.593, 'duration': 2.521}, {'end': 3041.724, 'text': 'First thing to think about, we just do some counting again.', 'start': 3039.723, 'duration': 2.001}, {'end': 3042.804, 'text': "We've got one blue marble.", 'start': 3041.764, 'duration': 1.04}, {'end': 3048.305, 'text': "How many ways can each conjecture produce an observation of one blue marble? Well, if they're all white, zero.", 'start': 3043.324, 'duration': 4.981}], 'summary': 'Bayesian inference updates counts with new data.', 'duration': 26.44, 'max_score': 3021.865, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo43021865.jpg'}], 'start': 2255.712, 'title': 'Bayesian inference', 'summary': "Introduces bayesian inference using the analogy of cristoforo colombo's mistaken navigation with a small globe, and discusses the small world-large world distinction in bayesian inference, using the example of a bag of marbles to illustrate bayesian data analysis.", 'chapters': [{'end': 2464.032, 'start': 2255.712, 'title': 'Bayesian inference: small world analogy', 'summary': "Introduces bayesian inference using the analogy of cristoforo colombo's mistaken navigation with a small globe, highlighting the importance of considering uncertainties and assumptions in modeling.", 'duration': 208.32, 'highlights': ["The small world analogy of Cristoforo Colombo's mistaken navigation with a small globe emphasizes the need to consider uncertainties and assumptions in modeling, similar to the process of Bayesian inference.", "Cristoforo Colombo's use of a small globe to plan his journey and his mistaken navigation serves as an analogy for the limitations and uncertainties in modeling, highlighting the importance of acknowledging potential errors in predictions.", "The irony of Colombo's reliance on a small globe for navigation, despite the accurate knowledge of the Earth's size by ancient Greeks, emphasizes the significance of acknowledging potential uncertainties and errors in predictions.", "The narrative of Cristoforo Colombo's mistaken navigation due to reliance on a small globe serves as an analogy for the uncertainties and potential errors in Bayesian inference, highlighting the need to consider the limitations of modeling assumptions."]}, {'end': 3146.02, 'start': 2464.032, 'title': 'Bayesian inference in small and large worlds', 'summary': 'Discusses the small world-large world distinction in bayesian inference, using the example of a bag of marbles to illustrate bayesian data analysis, and explains how bayesian updating is used to integrate new evidence into the analysis.', 'duration': 681.988, 'highlights': ["The chapter introduces the small world-large world distinction in Bayesian inference, emphasizing the difference between the optimal small world of the model and the real world where no optimal procedures exist, as highlighted by Leonard Jimmy Savage's influential work in Bayesian statistics. Leonard Jimmy Savage's influential work in Bayesian statistics introduces the small world-large world distinction, emphasizing the difference between the optimal small world of the model and the real world where no optimal procedures exist.", 'The example of a bag of marbles is used to illustrate Bayesian data analysis, where the enumeration of all possible contents of the bag and Bayesian inference are used to determine the most plausible contents based on observed data. The example of a bag of marbles is used to illustrate Bayesian data analysis, where the enumeration of all possible contents of the bag and Bayesian inference are used to determine the most plausible contents based on observed data.', 'The concept of Bayesian updating is explained, demonstrating how new evidence can be integrated into the analysis by multiplying the previous counts with the new counts, leading to an updated posterior probability distribution. The concept of Bayesian updating is explained, demonstrating how new evidence can be integrated into the analysis by multiplying the previous counts with the new counts, leading to an updated posterior probability distribution.']}], 'duration': 890.308, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo42255712.jpg', 'highlights': ["The small world analogy of Cristoforo Colombo's mistaken navigation with a small globe emphasizes the need to consider uncertainties and assumptions in modeling, similar to the process of Bayesian inference.", "The irony of Colombo's reliance on a small globe for navigation, despite the accurate knowledge of the Earth's size by ancient Greeks, emphasizes the significance of acknowledging potential uncertainties and errors in predictions.", "The narrative of Cristoforo Colombo's mistaken navigation due to reliance on a small globe serves as an analogy for the uncertainties and potential errors in Bayesian inference, highlighting the need to consider the limitations of modeling assumptions.", "The chapter introduces the small world-large world distinction in Bayesian inference, emphasizing the difference between the optimal small world of the model and the real world where no optimal procedures exist, as highlighted by Leonard Jimmy Savage's influential work in Bayesian statistics.", 'The example of a bag of marbles is used to illustrate Bayesian data analysis, where the enumeration of all possible contents of the bag and Bayesian inference are used to determine the most plausible contents based on observed data.', 'The concept of Bayesian updating is explained, demonstrating how new evidence can be integrated into the analysis by multiplying the previous counts with the new counts, leading to an updated posterior probability distribution.']}, {'end': 3580.579, 'segs': [{'end': 3230.29, 'src': 'heatmap', 'start': 3147.761, 'weight': 0.736, 'content': [{'end': 3159.264, 'text': 'So say, you have a friend who works at the factory and they testify that blue marbles are quite rare actually and but every bag contains at least one.', 'start': 3147.761, 'duration': 11.503}, {'end': 3162.767, 'text': 'They have a manufacturing process at the factory that they always put one blue marble.', 'start': 3159.284, 'duration': 3.483}, {'end': 3166.89, 'text': "because when they sell these bags of marbles and then someone buys them and there's not any blue marble, then they get angry.", 'start': 3162.767, 'duration': 4.123}, {'end': 3170.752, 'text': "So they ensure that there's always one blue marble.", 'start': 3167.93, 'duration': 2.822}, {'end': 3182.461, 'text': 'And he tells you in particular the information that in the factory manufacturing process they have no bags that are always white and no bags that are always blue.', 'start': 3172.714, 'duration': 9.747}, {'end': 3189.708, 'text': 'but there are, For every one bag that has three blue marbles, there are two that have two and three that have one.', 'start': 3182.461, 'duration': 7.247}, {'end': 3193.694, 'text': "So blue marbles are relatively rare in the bags, but they're always present.", 'start': 3190.389, 'duration': 3.305}, {'end': 3200.554, 'text': 'So these are the ratios, right? Does this make sense? How do we use this? Well, you guessed it.', 'start': 3194.431, 'duration': 6.123}, {'end': 3201.714, 'text': 'You multiply.', 'start': 3200.794, 'duration': 0.92}, {'end': 3210.318, 'text': "It's the same because these are ways that bags can happen, right? There are three ways that you can get a bag with one blue marble.", 'start': 3202.214, 'duration': 8.104}, {'end': 3212.999, 'text': 'For every two ways, you can get a bag with two blue marbles.', 'start': 3211.018, 'duration': 1.981}, {'end': 3215.1, 'text': 'For every one way, you can get a bag with only one blue marble.', 'start': 3213.019, 'duration': 2.081}, {'end': 3218.161, 'text': 'So these are just ways, and we just add them into our prior ways.', 'start': 3215.48, 'duration': 2.681}, {'end': 3223.083, 'text': 'So we had three prior ways from our previous four drawn marbles from the bag.', 'start': 3218.201, 'duration': 4.882}, {'end': 3226.344, 'text': 'that it could have been one blue marble in the bag.', 'start': 3223.7, 'duration': 2.644}, {'end': 3230.29, 'text': "We multiply this by the factory count of three, and now this seems more plausible because it's none.", 'start': 3226.745, 'duration': 3.545}], 'summary': 'Blue marbles are relatively rare in the bags, but always present. the manufacturing process ensures at least one blue marble in every bag.', 'duration': 82.529, 'max_score': 3147.761, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo43147761.jpg'}, {'end': 3212.999, 'src': 'embed', 'start': 3167.93, 'weight': 0, 'content': [{'end': 3170.752, 'text': "So they ensure that there's always one blue marble.", 'start': 3167.93, 'duration': 2.822}, {'end': 3182.461, 'text': 'And he tells you in particular the information that in the factory manufacturing process they have no bags that are always white and no bags that are always blue.', 'start': 3172.714, 'duration': 9.747}, {'end': 3189.708, 'text': 'but there are, For every one bag that has three blue marbles, there are two that have two and three that have one.', 'start': 3182.461, 'duration': 7.247}, {'end': 3193.694, 'text': "So blue marbles are relatively rare in the bags, but they're always present.", 'start': 3190.389, 'duration': 3.305}, {'end': 3200.554, 'text': 'So these are the ratios, right? Does this make sense? How do we use this? Well, you guessed it.', 'start': 3194.431, 'duration': 6.123}, {'end': 3201.714, 'text': 'You multiply.', 'start': 3200.794, 'duration': 0.92}, {'end': 3210.318, 'text': "It's the same because these are ways that bags can happen, right? There are three ways that you can get a bag with one blue marble.", 'start': 3202.214, 'duration': 8.104}, {'end': 3212.999, 'text': 'For every two ways, you can get a bag with two blue marbles.', 'start': 3211.018, 'duration': 1.981}], 'summary': 'Bags contain varying amounts of blue marbles, with a ratio of 1:2:3, and they always contain at least one blue marble.', 'duration': 45.069, 'max_score': 3167.93, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo43167930.jpg'}, {'end': 3374.009, 'src': 'heatmap', 'start': 3260.125, 'weight': 1, 'content': [{'end': 3264.346, 'text': "You're not going to literally hand count possibilities in your model, of course.", 'start': 3260.125, 'duration': 4.221}, {'end': 3265.687, 'text': "We're going to use the computer to do that.", 'start': 3264.366, 'duration': 1.321}, {'end': 3271.895, 'text': "And what's even more amazing about it, is your computer is going to count an infinite number of different conjectures.", 'start': 3266.167, 'duration': 5.728}, {'end': 3275.518, 'text': "It's going to rank them all and count up all the different ways they could produce the data.", 'start': 3272.136, 'duration': 3.382}, {'end': 3283.284, 'text': 'And this is easy thanks to a magical invention called calculus, which is invented near here, actually, by a fellow named Leibniz.', 'start': 3275.558, 'duration': 7.726}, {'end': 3287.126, 'text': "And there's also that Newton guy, but we won't talk about him.", 'start': 3283.924, 'duration': 3.202}, {'end': 3293.531, 'text': 'So what we end up with is these counts get converted to plausibilities.', 'start': 3288.667, 'duration': 4.864}, {'end': 3296.879, 'text': 'Because these counts are going to get big very, very fast.', 'start': 3294.198, 'duration': 2.681}, {'end': 3300.3, 'text': 'If you have a reasonably sized data set, there are going to be a really,', 'start': 3297.079, 'duration': 3.221}, {'end': 3307.102, 'text': 'really large number of different possible paths through the garden of working data that could produce any particular data set.', 'start': 3300.3, 'duration': 6.802}, {'end': 3308.622, 'text': "That's how combinatorics works.", 'start': 3307.182, 'duration': 1.44}, {'end': 3311.383, 'text': 'Combinatoric counts get big really, really fast.', 'start': 3309.242, 'duration': 2.141}, {'end': 3313.143, 'text': "You don't want to carry those sums around.", 'start': 3311.723, 'duration': 1.42}, {'end': 3318.565, 'text': 'So instead, we normalize the sums so that there are numbers between 0 and 1.', 'start': 3313.763, 'duration': 4.802}, {'end': 3319.665, 'text': 'And now there are probabilities.', 'start': 3318.565, 'duration': 1.1}, {'end': 3321.645, 'text': "And that's what probability theory is.", 'start': 3320.085, 'duration': 1.56}, {'end': 3323.146, 'text': "It's normalized counting.", 'start': 3321.826, 'duration': 1.32}, {'end': 3325.987, 'text': "And it's wonderful.", 'start': 3325.287, 'duration': 0.7}, {'end': 3334.809, 'text': 'But then all of the actions of probability theory are just ways of dealing with counts that have been normalized to be between 0 and 1.', 'start': 3328.528, 'duration': 6.281}, {'end': 3337.43, 'text': 'And you can derive it all that way.', 'start': 3334.809, 'duration': 2.621}, {'end': 3338.89, 'text': 'And this is not mysterious.', 'start': 3337.99, 'duration': 0.9}, {'end': 3339.85, 'text': 'Mathematicians know this.', 'start': 3338.95, 'duration': 0.9}, {'end': 3342.311, 'text': 'But then they get the action to take on their own flavor.', 'start': 3340.351, 'duration': 1.96}, {'end': 3343.651, 'text': 'And they look more mysterious.', 'start': 3342.631, 'duration': 1.02}, {'end': 3346.152, 'text': 'But this makes it convenient, actually.', 'start': 3344.512, 'duration': 1.64}, {'end': 3351.973, 'text': "It's nicer to work with these sums that are between 0 and 1.", 'start': 3346.192, 'duration': 5.781}, {'end': 3354.814, 'text': 'So these plausibilities end up getting converted in the right column.', 'start': 3351.973, 'duration': 2.841}, {'end': 3358.607, 'text': 'so that all of them add up to one.', 'start': 3356.744, 'duration': 1.863}, {'end': 3359.469, 'text': 'How do you do that??', 'start': 3358.868, 'duration': 0.601}, {'end': 3366.041, 'text': 'You just sum up all the thing, all those numbers in the middle column, and then divide each by that sum, and you get a plausibility on the right.', 'start': 3359.549, 'duration': 6.492}, {'end': 3374.009, 'text': "Usually, when we work in statistical models, we assign some parameter to index the different possibilities of what's in the bag.", 'start': 3368.387, 'duration': 5.622}], 'summary': 'Using calculus, we convert combinatoric counts to probabilities, and normalize them between 0 and 1 for statistical models.', 'duration': 113.884, 'max_score': 3260.125, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo43260125.jpg'}, {'end': 3300.3, 'src': 'embed', 'start': 3272.136, 'weight': 4, 'content': [{'end': 3275.518, 'text': "It's going to rank them all and count up all the different ways they could produce the data.", 'start': 3272.136, 'duration': 3.382}, {'end': 3283.284, 'text': 'And this is easy thanks to a magical invention called calculus, which is invented near here, actually, by a fellow named Leibniz.', 'start': 3275.558, 'duration': 7.726}, {'end': 3287.126, 'text': "And there's also that Newton guy, but we won't talk about him.", 'start': 3283.924, 'duration': 3.202}, {'end': 3293.531, 'text': 'So what we end up with is these counts get converted to plausibilities.', 'start': 3288.667, 'duration': 4.864}, {'end': 3296.879, 'text': 'Because these counts are going to get big very, very fast.', 'start': 3294.198, 'duration': 2.681}, {'end': 3300.3, 'text': 'If you have a reasonably sized data set, there are going to be a really,', 'start': 3297.079, 'duration': 3.221}], 'summary': 'Calculus enables counting and ranking data efficiently.', 'duration': 28.164, 'max_score': 3272.136, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo43272136.jpg'}, {'end': 3480.571, 'src': 'heatmap', 'start': 3438.668, 'weight': 0.761, 'content': [{'end': 3443.213, 'text': 'So plausibility is probability in this view of it.', 'start': 3438.668, 'duration': 4.545}, {'end': 3447.238, 'text': "It's just subset of non-negative real numbers that sum to one.", 'start': 3444.437, 'duration': 2.801}, {'end': 3453.42, 'text': 'And those are the relative number of ways that each of these conjectures could be true, conditional on the evidence.', 'start': 3447.558, 'duration': 5.862}, {'end': 3456.622, 'text': 'I will say phrases like that a thousand times in this course.', 'start': 3454.081, 'duration': 2.541}, {'end': 3460.363, 'text': 'It will become second nature to you to think of it that way.', 'start': 3458.342, 'duration': 2.021}, {'end': 3465.625, 'text': 'And probability theory is just shortcuts for counting.', 'start': 3463.384, 'duration': 2.241}, {'end': 3467.106, 'text': 'So Bayesian inference is just counting.', 'start': 3465.845, 'duration': 1.261}, {'end': 3470.267, 'text': "It's counting weird things, but it's counting.", 'start': 3468.666, 'duration': 1.601}, {'end': 3477.99, 'text': "Okay, so I got one more minute, but I want to prep where we're going to start next time a bit,", 'start': 3473.047, 'duration': 4.943}, {'end': 3480.571, 'text': "so that I can start here again and you feel like there's some continuity.", 'start': 3477.99, 'duration': 2.581}], 'summary': 'Bayesian inference is about probabilities and counting.', 'duration': 41.903, 'max_score': 3438.668, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo43438668.jpg'}, {'end': 3546.911, 'src': 'embed', 'start': 3522.777, 'weight': 3, 'content': [{'end': 3529.166, 'text': 'If we blind ourselves logically to the exact numbers in the spreadsheet, only knowing about the generative process,', 'start': 3522.777, 'duration': 6.389}, {'end': 3530.586, 'text': 'the nature of the scientific background?', 'start': 3529.166, 'duration': 1.42}, {'end': 3533.247, 'text': "What can we design a model and that's what we want to do.", 'start': 3530.986, 'duration': 2.261}, {'end': 3538.029, 'text': 'Then we condition on the data using Bayesian inference.', 'start': 3534.567, 'duration': 3.462}, {'end': 3538.609, 'text': 'We update.', 'start': 3538.069, 'duration': 0.54}, {'end': 3543.17, 'text': "We've got all the conjectures that were nominated for the scientific information and then we count ways.", 'start': 3538.629, 'duration': 4.541}, {'end': 3546.911, 'text': 'And then we get critical.', 'start': 3545.591, 'duration': 1.32}], 'summary': 'Using bayesian inference to update and generate scientific models.', 'duration': 24.134, 'max_score': 3522.777, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo43522777.jpg'}], 'start': 3147.761, 'title': 'Probability and bayesian models', 'summary': "Discusses a factory's manufacturing process with ratios of 3:2:1 for bags containing blue marbles and emphasizes using multiplication to calculate likelihood, and explains the mechanics of bayesian models, involving calculus, normalization to probabilities, and bayesian inference.", 'chapters': [{'end': 3252.604, 'start': 3147.761, 'title': 'Probability and manufacturing process', 'summary': "Discusses a factory's manufacturing process where blue marbles are relatively rare in the bags but always present, with ratios of 3:2:1 for bags containing three, two, and one blue marbles respectively, and emphasizes using multiplication to calculate the likelihood of different bag compositions.", 'duration': 104.843, 'highlights': ['Bags always contain at least one blue marble to avoid customer dissatisfaction, with a manufacturing process ensuring the presence of a blue marble in every bag. The factory ensures that every bag contains at least one blue marble to prevent customer dissatisfaction, which is achieved through their manufacturing process.', 'Ratios of bags with three, two, and one blue marbles are 3:2:1, indicating that blue marbles are relatively rare in the bags but always present. The ratios of bags with three, two, and one blue marbles are 3:2:1, signifying that while blue marbles are relatively rare in the bags, they are always present.', 'Using multiplication to calculate the likelihood of different bag compositions based on the manufacturing process and prior ways of obtaining blue marbles. The chapter emphasizes using multiplication to calculate the likelihood of different bag compositions based on the manufacturing process and prior ways of obtaining blue marbles from the bags.']}, {'end': 3580.579, 'start': 3254.504, 'title': 'Understanding bayesian models', 'summary': 'Explains the mechanics of bayesian models, involving the use of calculus to count and rank an infinite number of conjectures and normalize counts to probabilities, leading to bayesian inference and a recursive cycle of designing, updating, and critically evaluating models.', 'duration': 326.075, 'highlights': ['Bayesian models involve using calculus to count and rank an infinite number of conjectures and normalize counts to probabilities. The use of calculus enables the computer to count an infinite number of different conjectures and rank them, while normalizing the counts to probabilities between 0 and 1.', 'The process of Bayesian inference involves designing a model, updating based on the data using Bayesian inference, and evaluating and potentially changing the model in a recursive cycle. The modeling exercise in Bayesian inference follows a recursive cycle of three steps: designing the model based on scientific information, updating based on the data using Bayesian inference, and critically evaluating and potentially changing the model based on any missing or incorrect factors.', 'The chapter emphasizes the recursive cycle of designing, updating, and critically evaluating Bayesian models transparently and publicly to seek input and assistance from peers. The process of designing, updating, and critically evaluating Bayesian models is intended to be transparent and public, allowing peers to contribute and provide assistance in refining the models.']}], 'duration': 432.818, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/4WVelCswXo4/pics/4WVelCswXo43147761.jpg', 'highlights': ['The ratios of bags with three, two, and one blue marbles are 3:2:1, signifying that while blue marbles are relatively rare in the bags, they are always present.', 'The factory ensures that every bag contains at least one blue marble to prevent customer dissatisfaction, which is achieved through their manufacturing process.', 'Using multiplication to calculate the likelihood of different bag compositions based on the manufacturing process and prior ways of obtaining blue marbles from the bags.', 'The process of Bayesian inference involves designing a model, updating based on the data using Bayesian inference, and evaluating and potentially changing the model in a recursive cycle.', 'The use of calculus enables the computer to count an infinite number of different conjectures and rank them, while normalizing the counts to probabilities between 0 and 1.', 'The process of designing, updating, and critically evaluating Bayesian models is intended to be transparent and public, allowing peers to contribute and provide assistance in refining the models.']}], 'highlights': ['The course provides a structured 10-week curriculum with 20 lectures and 2 hours of teaching per week, along with supplementary materials such as a book and software.', 'The focus of the course is on delivering practical model building and criticizing skills, primarily through coding exercises rather than analytical mathematics.', 'The chapter emphasizes the limitations of statistical tests in falsifying hypotheses and highlights the need to view statistical models as tools rather than rational entities.', 'The BUGS project revolutionized the use of Bayesian methods in stats departments and private research.', "The small world analogy of Cristoforo Colombo's mistaken navigation with a small globe emphasizes the need to consider uncertainties and assumptions in modeling, similar to the process of Bayesian inference.", 'The concept of Bayesian updating is explained, demonstrating how new evidence can be integrated into the analysis by multiplying the previous counts with the new counts, leading to an updated posterior probability distribution.']}