title
Customer churn prediction using ANN | Deep Learning Tutorial 18 (Tensorflow2.0, Keras & Python)

description
In this video we will build a customer churn prediction model using artificial neural network or ANN. Customer churn measures how and why are customers leaving the business. WE will use telecom customer churn dataset from kaggle (link below) and build a deep learning model for churn prediction. We will also understand precision,recalll and accuracy of this model by using confusion matrix and classification report #customerchurnprediction #customerchurn #customerchurnpredictionusingANN #ANN #deeplearningANN #deeplearningtutorial Do you want to learn technology from me? Check https://codebasics.io/?utm_source=description&utm_medium=yt&utm_campaign=description&utm_id=description for my affordable video courses. Code: https://github.com/codebasics/deep-learning-keras-tf-tutorial/blob/master/11_chrun_prediction/churn.ipynb Exercise: For exercise go at the end of above notebook. There is a link to kaggle dataset and exercise description Dataset: https://www.kaggle.com/blastchar/telco-customer-churn Next video: https://www.youtube.com/watch?v=2osIZ-dSPGE&list=PLeo1K3hjS3uu7CxAacxVndI4bE_o3BDtO&index=19 Previous video: https://www.youtube.com/watch?v=YmDaqXMIoeY&list=PLeo1K3hjS3uu7CxAacxVndI4bE_o3BDtO&index=17 Deep learning playlist: https://www.youtube.com/playlist?list=PLeo1K3hjS3uu7CxAacxVndI4bE_o3BDtO Machine learning playlist : https://www.youtube.com/playlist?list=PLeo1K3hjS3uvCeTYTeyfe0-rN5r8zn9rw #️⃣ Social Media #️⃣ 🔗 Discord: https://discord.gg/r42Kbuk 📸 Dhaval's Personal Instagram: https://www.instagram.com/dhavalsays/ 📸 Instagram: https://www.instagram.com/codebasicshub/ 🔊 Facebook: https://www.facebook.com/codebasicshub 📝 Linkedin (Personal): https://www.linkedin.com/in/dhavalsays/ 📝 Linkedin (Codebasics): https://www.linkedin.com/company/codebasics/ 📱 Twitter: https://twitter.com/codebasicshub 🔗 Patreon: https://www.patreon.com/codebasics?fan_landing=true DISCLAIMER: All opinions expressed in this video are of my own and not that of my employers'.

detail
{'title': 'Customer churn prediction using ANN | Deep Learning Tutorial 18 (Tensorflow2.0, Keras & Python)', 'heatmap': [{'end': 295.057, 'start': 266.069, 'weight': 0.711}, {'end': 884.72, 'start': 829.403, 'weight': 0.731}, {'end': 1222.682, 'start': 1187.774, 'weight': 0.867}, {'end': 1617.22, 'start': 1557.456, 'weight': 0.709}, {'end': 1763.193, 'start': 1708.813, 'weight': 0.85}, {'end': 1831.527, 'start': 1806.158, 'weight': 0.743}, {'end': 2105.381, 'start': 2070.945, 'weight': 1}], 'summary': 'Discusses using neural networks to predict customer churn, emphasizing the importance of retaining customers and demonstrating model evaluation with 78% accuracy and precision of 0.83 for leaving customers and 0.63 for staying customers.', 'chapters': [{'end': 143.514, 'segs': [{'end': 92.292, 'src': 'embed', 'start': 41.696, 'weight': 0, 'content': [{'end': 49.724, 'text': 'and deep learning can help you with a great amount in you know, kind of measuring why customers are leaving and if,', 'start': 41.696, 'duration': 8.028}, {'end': 56.693, 'text': "once you know why customers are living, you can take appropriate business actions so that the customers don't leave the business,", 'start': 49.724, 'duration': 6.969}, {'end': 65.922, 'text': "we will write code in python using tensorflow, we'll be building simple artificial neural network And we will also discuss about precision recall,", 'start': 56.693, 'duration': 9.229}, {'end': 67.983, 'text': 'F1 score all of those terms.', 'start': 65.922, 'duration': 2.061}, {'end': 69.144, 'text': 'We will demystify.', 'start': 68.083, 'duration': 1.061}, {'end': 72.125, 'text': 'We will also run a classification report.', 'start': 69.164, 'duration': 2.961}, {'end': 76.547, 'text': 'So overall, it will be a very useful hands-on coding session today.', 'start': 72.586, 'duration': 3.961}, {'end': 80.649, 'text': 'In the end, we will also have an exercise for you to solve.', 'start': 77.188, 'duration': 3.461}, {'end': 81.77, 'text': "So let's begin.", 'start': 80.709, 'duration': 1.061}, {'end': 88.187, 'text': 'I will be using Telecom Customer Churn Dataset from Kaggle.', 'start': 83.621, 'duration': 4.566}, {'end': 92.292, 'text': 'I have a link of this dataset in the video description below.', 'start': 88.607, 'duration': 3.685}], 'summary': 'Using deep learning to analyze customer churn with python and tensorflow, including precision, recall, and f1 score. hands-on session with telecom customer churn dataset from kaggle.', 'duration': 50.596, 'max_score': 41.696, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U41696.jpg'}], 'start': 0.349, 'title': 'Customer churn prediction', 'summary': "Discusses the use of neural networks to predict customer churn, emphasizing its importance in retaining customers. it also mentions the use of precision, recall, and f1 score in evaluating the model's performance, and will be demonstrated using the telecom customer churn dataset from kaggle.", 'chapters': [{'end': 143.514, 'start': 0.349, 'title': 'Customer churn prediction using neural network', 'summary': "Discusses the use of artificial neural networks to predict customer churn, emphasizing its importance in retaining customers and the practical application in various industries. it also mentions the use of precision, recall, and f1 score in evaluating the model's performance and will be demonstrated using the telecom customer churn dataset from kaggle.", 'duration': 143.165, 'highlights': ['Artificial neural networks can help measure why customers are leaving, enabling businesses to take appropriate actions to retain them.', "Discussion on precision, recall, and F1 score for evaluating the model's performance.", 'Utilizing the Telecom Customer Churn Dataset from Kaggle for hands-on coding and practical exercises.']}], 'duration': 143.165, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U349.jpg', 'highlights': ['Artificial neural networks can help measure why customers are leaving, enabling businesses to take appropriate actions to retain them.', 'Utilizing the Telecom Customer Churn Dataset from Kaggle for hands-on coding and practical exercises.', "Discussion on precision, recall, and F1 score for evaluating the model's performance."]}, {'end': 513.717, 'segs': [{'end': 174.733, 'src': 'embed', 'start': 143.514, 'weight': 4, 'content': [{'end': 145.875, 'text': 'then maybe customer might decide to leave.', 'start': 143.514, 'duration': 2.361}, {'end': 152.791, 'text': 'I have loaded that data in my Jupyter notebook and my data frame looks something like this', 'start': 147.845, 'duration': 4.946}, {'end': 156.796, 'text': 'I have also imported some useful libraries.', 'start': 152.811, 'duration': 3.985}, {'end': 165.426, 'text': 'And now, whenever you are working on a machine learning problem, the first thing that you do is you try to do data exploration.', 'start': 157.837, 'duration': 7.589}, {'end': 170.65, 'text': 'Now, the very first thing I noticed was customer ID is useless.', 'start': 166.227, 'duration': 4.423}, {'end': 174.733, 'text': "When you're building machine learning model, customer ID is not going to help you.", 'start': 171.49, 'duration': 3.243}], 'summary': 'Customer id is useless in machine learning for customer retention.', 'duration': 31.219, 'max_score': 143.514, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U143514.jpg'}, {'end': 260.887, 'src': 'embed', 'start': 199.457, 'weight': 0, 'content': [{'end': 202.902, 'text': 'Also for remaining columns it is showing the data type.', 'start': 199.457, 'duration': 3.445}, {'end': 211.444, 'text': 'One quick thing I noticed was my total charges here is object.', 'start': 204.321, 'duration': 7.123}, {'end': 216.065, 'text': 'See my monthly charges is float, but total charges is object.', 'start': 212.544, 'duration': 3.521}, {'end': 224.208, 'text': "So why is that? So what's really going on here? So here what I will do is DF dot total charges.", 'start': 216.125, 'duration': 8.083}, {'end': 230.866, 'text': 'values. you see, these are string.', 'start': 227.803, 'duration': 3.063}, {'end': 233.589, 'text': 'actually, I need to convert it to a number.', 'start': 230.866, 'duration': 2.723}, {'end': 243.719, 'text': "I don't know why the data is like this if you look at monthly charges, and then monthly charges will be a number.", 'start': 233.589, 'duration': 10.13}, {'end': 255.526, 'text': 'so first thing, of course, you need to convert this into a number column, and the way you do that is by using PD to numeric function,', 'start': 243.719, 'duration': 11.807}, {'end': 256.486, 'text': 'something like this.', 'start': 255.526, 'duration': 0.96}, {'end': 260.887, 'text': 'so when you do that, it will convert this into numbers.', 'start': 256.486, 'duration': 4.401}], 'summary': 'The total charges column needs to be converted to numbers using the pd to numeric function.', 'duration': 61.43, 'max_score': 199.457, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U199457.jpg'}, {'end': 295.057, 'src': 'heatmap', 'start': 266.069, 'weight': 0.711, 'content': [{'end': 268.529, 'text': 'so we need to tackle the spaces.', 'start': 266.069, 'duration': 2.46}, {'end': 273.762, 'text': 'so first i want to just quickly look at those rows.', 'start': 268.529, 'duration': 5.233}, {'end': 275.123, 'text': 'which has space?', 'start': 273.762, 'duration': 1.361}, {'end': 277.845, 'text': "i want to see what's going on.", 'start': 275.123, 'duration': 2.722}, {'end': 285.37, 'text': "so one way to do that would be doing this, which is, i'm doing pd to numeric.", 'start': 277.845, 'duration': 7.525}, {'end': 295.057, 'text': 'and, by the way, when you do pd to numeric and when you supply errors is equal to cores, it will ignore the errors.', 'start': 285.37, 'duration': 9.687}], 'summary': 'Identifying and handling spaces in rows using pd to numeric and ignoring errors.', 'duration': 28.988, 'max_score': 266.069, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U266069.jpg'}, {'end': 368.216, 'src': 'embed', 'start': 334.619, 'weight': 3, 'content': [{'end': 344.903, 'text': 'So when you have a data frame like this, and when you supply this whole thing, it will serve as an index and wherever the value is set to true,', 'start': 334.619, 'duration': 10.284}, {'end': 346.584, 'text': 'it will show you that row.', 'start': 344.903, 'duration': 1.681}, {'end': 359.73, 'text': 'So see I found total these many rows and if you scroll here you will see the total charges is blank for all these rows.', 'start': 347.505, 'duration': 12.225}, {'end': 368.216, 'text': 'now i can drop these rows because if you look at this data frame, i have like what 11 rows here.', 'start': 360.39, 'duration': 7.826}], 'summary': 'Using the data frame as an index, locate and drop 11 rows with blank total charges.', 'duration': 33.597, 'max_score': 334.619, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U334619.jpg'}, {'end': 460.309, 'src': 'embed', 'start': 430.168, 'weight': 1, 'content': [{'end': 437.153, 'text': "Alright, so let's drop all of these rows which is close to 11.", 'start': 430.168, 'duration': 6.985}, {'end': 437.513, 'text': 'I think 11, yeah.', 'start': 437.153, 'duration': 0.36}, {'end': 441.991, 'text': 'So how do you drop this rows? Well, you can do this.', 'start': 439.349, 'duration': 2.642}, {'end': 448.014, 'text': 'You can set total charges if it is not equal to space, then keep it, otherwise drop it.', 'start': 442.631, 'duration': 5.383}, {'end': 456.24, 'text': 'And the new data frame, this will return data frame, by the way, and I want to store it into a new data frame called DF1.', 'start': 448.735, 'duration': 7.505}, {'end': 460.309, 'text': 'hope it is all making sense.', 'start': 458.487, 'duration': 1.822}], 'summary': 'Dropped rows close to 11, stored into new data frame df1.', 'duration': 30.141, 'max_score': 430.168, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U430168.jpg'}], 'start': 143.514, 'title': 'Data cleaning and null values handling', 'summary': 'Covers data exploration, cleaning, and null value handling in a machine learning project, including dropping irrelevant columns, converting data types, and handling errors. it also explains how to identify and drop null value rows in a pandas dataframe, dropping 11 rows out of 7,000, and using iloc to access specific row values, and applying a condition to drop rows based on a specific column, resulting in the removal of all blank values in the total charges column.', 'chapters': [{'end': 309.485, 'start': 143.514, 'title': 'Data exploration and cleaning', 'summary': 'Covers data exploration and cleaning in a machine learning project, including dropping irrelevant columns, converting data types, and handling errors, with a focus on customer id and total charges.', 'duration': 165.971, 'highlights': ['The first step in working on a machine learning problem is data exploration, where the useless column like customer ID is dropped in order to clean and update the data frame.', "Identifying and addressing discrepancies in data types, such as converting 'total charges' from object to numeric format, is crucial for accurate analysis and model building.", "Utilizing the 'PD to_numeric' function to convert data columns into numbers, and handling errors by ignoring spaces and replacing them with 'n a', is a key aspect of data cleaning and preparation."]}, {'end': 513.717, 'start': 309.485, 'title': 'Handling null values in pandas', 'summary': 'Explains how to identify and drop null value rows in a pandas dataframe, dropping 11 rows out of 7,000, and using iloc to access specific row values, and applying a condition to drop rows based on a specific column, resulting in the removal of all blank values in the total charges column.', 'duration': 204.232, 'highlights': ['The chapter demonstrates how to drop 11 rows out of 7,000 in a pandas dataframe, resulting in the removal of all blank values in the total charges column.', 'Using iloc to access the specific row (488) and identifying the blank value in the total charges column.', 'Explanation of applying a condition to drop rows based on a specific column, resulting in the removal of all blank values in the total charges column.', 'The tutorial playlist for pandas on youtube is mentioned, catering to a wide audience including high school students.']}], 'duration': 370.203, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U143514.jpg', 'highlights': ["Converting 'total charges' from object to numeric format is crucial for accurate analysis", "Dropping 11 rows out of 7,000 in a pandas dataframe removes all blank values in 'total charges'", "Utilizing 'PD to_numeric' function to convert data columns into numbers is key for data cleaning", "Applying a condition to drop rows based on a specific column removes all blank values in 'total charges'", 'Useless columns like customer ID are dropped in data exploration to clean and update the data frame']}, {'end': 869.267, 'segs': [{'end': 869.267, 'src': 'embed', 'start': 781.569, 'weight': 0, 'content': [{'end': 788.215, 'text': 'if you observe this carefully, you will notice the people who are with the company for a long time.', 'start': 781.569, 'duration': 6.646}, {'end': 793.941, 'text': "let's say, 70 months here is like 70 months there more.", 'start': 788.215, 'duration': 5.726}, {'end': 795.602, 'text': 'majority of the customers are not living.', 'start': 793.941, 'duration': 1.661}, {'end': 809.335, 'text': 'see around more than thousand customers with tenure equal to 70 or more are not living and Less than maybe less than 100 customers are leaving who have that kind of tenure.', 'start': 795.602, 'duration': 13.733}, {'end': 818.059, 'text': "So see this kind of visualization can help you give a quick insights on what's going on with your data.", 'start': 809.935, 'duration': 8.124}, {'end': 822.12, 'text': 'I did this plot for tenure.', 'start': 819.079, 'duration': 3.041}, {'end': 826.722, 'text': "We can do the same plot for, let's say, monthly charges.", 'start': 822.34, 'duration': 4.382}, {'end': 829.343, 'text': 'You know, sometimes the monthly charge is very high.', 'start': 827.442, 'duration': 1.901}, {'end': 830.284, 'text': 'Customers might leave.', 'start': 829.403, 'duration': 0.881}, {'end': 834.386, 'text': 'So all you will do it exactly same chart.', 'start': 831.264, 'duration': 3.122}, {'end': 838.388, 'text': "I'm doing copy paste of code just to save time.", 'start': 835.286, 'duration': 3.102}, {'end': 842.731, 'text': "But here really it's the same code as previous one.", 'start': 838.968, 'duration': 3.763}, {'end': 846.052, 'text': 'But instead of tenure, I have monthly charges.', 'start': 843.251, 'duration': 2.801}, {'end': 848.555, 'text': "Okay, it's the same chart.", 'start': 846.833, 'duration': 1.722}, {'end': 859.861, 'text': 'just the column is different and here what it is saying is our customers, who have very high charge, for example, close to 120.', 'start': 848.555, 'duration': 11.306}, {'end': 863.023, 'text': "see, they're there, they are living more.", 'start': 859.861, 'duration': 3.162}, {'end': 864.364, 'text': 'customers were in mid-range.', 'start': 863.023, 'duration': 1.341}, {'end': 866.906, 'text': 'they are kind of okay.', 'start': 864.364, 'duration': 2.542}, {'end': 869.267, 'text': 'overall, this company has customer living.', 'start': 866.906, 'duration': 2.361}], 'summary': 'Visualization reveals majority of long-tenured customers are living, with over 1000 customers at 70+ months tenure, while less than 100 are leaving.', 'duration': 87.698, 'max_score': 781.569, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U781569.jpg'}], 'start': 513.717, 'title': 'Visualizing customer churn', 'summary': 'Discusses techniques for visualizing customer churn, such as using histograms and charts to identify loyal customers, predict churn likelihood, and analyze tenure and monthly charges for insights, such as the impact on customer retention and high churn rate after 70 months.', 'chapters': [{'end': 755.045, 'start': 513.717, 'title': 'Data visualization for customer churn', 'summary': 'Discusses visualizing customer churn using histogram where loyal customers are identified based on tenure and their likelihood of leaving is visualized using green and red colors.', 'duration': 241.328, 'highlights': ["Loyal customers' tenure is visualized to identify the likelihood of them leaving, with green indicating they are staying and red indicating they are leaving.", 'The chapter discusses the process of visualizing customer churn using a histogram, with a focus on identifying loyal customers based on their tenure with the company.', 'The process of identifying and visualizing the tenure of customers who are leaving and those who are staying is explained in the context of customer churn analysis.']}, {'end': 869.267, 'start': 755.045, 'title': 'Customer churn visualization', 'summary': 'Discusses customer churn prediction visualization using a chart to analyze tenure and monthly charges, revealing insights such as the high number of customers leaving after 70 months and the impact of monthly charges on customer retention.', 'duration': 114.222, 'highlights': ['The majority of the customers are not living after 70 months, with over 1000 customers with tenure equal to 70 or more not living.', 'Customers with very high monthly charges, close to 120, are more likely to stay, while those in the mid-range are relatively okay.', 'The visualization provides quick insights into the data, enabling analysis of customer behavior based on tenure and monthly charges.']}], 'duration': 355.55, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U513717.jpg', 'highlights': ['Over 1000 customers have a tenure of 70 months or more and are not staying.', 'Visualization enables quick insights into customer behavior based on tenure and monthly charges.', 'Customers with monthly charges close to 120 are more likely to stay.']}, {'end': 1915.379, 'segs': [{'end': 900.391, 'src': 'embed', 'start': 869.267, 'weight': 3, 'content': [{'end': 873.129, 'text': 'you see, like customer with no is more than customer with yes.', 'start': 869.267, 'duration': 3.862}, {'end': 876.852, 'text': 'so because overall company is in inner trouble,', 'start': 873.129, 'duration': 3.723}, {'end': 884.72, 'text': 'I would say One thing I noticed while I was looking at a data frame is many of the columns has yes and no.', 'start': 876.852, 'duration': 7.868}, {'end': 894.749, 'text': 'So I want to find out the unique values in each column and kind of figure out the yes, no column so that I can do label encoding.', 'start': 886.226, 'duration': 8.523}, {'end': 900.391, 'text': 'And for that, I will quickly run a for loop.', 'start': 896.069, 'duration': 4.322}], 'summary': "Identified need for label encoding due to high frequency of 'yes' and 'no' values in many columns.", 'duration': 31.124, 'max_score': 869.267, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U869267.jpg'}, {'end': 997.344, 'src': 'embed', 'start': 970.678, 'weight': 4, 'content': [{'end': 976.363, 'text': "Now I want to put this code in the function because we'll be using it a lot.", 'start': 970.678, 'duration': 5.685}, {'end': 982.968, 'text': 'And I also want to print those columns where the data type is an object.', 'start': 977.183, 'duration': 5.785}, {'end': 987.115, 'text': 'Because those are categorical columns like tenure, etc.', 'start': 984.312, 'duration': 2.803}, {'end': 987.995, 'text': 'are numerical columns.', 'start': 987.135, 'duration': 0.86}, {'end': 990.197, 'text': 'So I want to skip it and this is the way you skip it.', 'start': 988.015, 'duration': 2.182}, {'end': 994.521, 'text': 'So now it is showing you all only the object type of column.', 'start': 991.078, 'duration': 3.443}, {'end': 997.344, 'text': 'So let me put this into a function.', 'start': 994.561, 'duration': 2.783}], 'summary': 'Creating a function to print categorical columns.', 'duration': 26.666, 'max_score': 970.678, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U970678.jpg'}, {'end': 1066.284, 'src': 'embed', 'start': 1033.8, 'weight': 7, 'content': [{'end': 1036.421, 'text': 'If you have no phone service, I want to replace it with no.', 'start': 1033.8, 'duration': 2.621}, {'end': 1042.143, 'text': 'OK And so let me just see here.', 'start': 1036.861, 'duration': 5.282}, {'end': 1045.674, 'text': 'For example here, no internet service.', 'start': 1043.53, 'duration': 2.144}, {'end': 1052.406, 'text': 'So in this row, instead of this, I want to say no, because that is the same as this no.', 'start': 1046.155, 'duration': 6.251}, {'end': 1062.582, 'text': 'And you all know if you follow my pandas tutorial that data frame has a function called replace.', 'start': 1057.239, 'duration': 5.343}, {'end': 1066.284, 'text': 'So you can replace no intended service with no.', 'start': 1063.382, 'duration': 2.902}], 'summary': "Replace no phone service with no using pandas' replace function.", 'duration': 32.484, 'max_score': 1033.8, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1033800.jpg'}, {'end': 1148.92, 'src': 'embed', 'start': 1119.003, 'weight': 6, 'content': [{'end': 1125.705, 'text': 'But first, what I want to do is I want to replace yes and no with maybe one and zero.', 'start': 1119.003, 'duration': 6.702}, {'end': 1132.327, 'text': 'Right Makes sense because we all know machine learning models do not understand text.', 'start': 1126.745, 'duration': 5.582}, {'end': 1137.669, 'text': 'So we have to convert every text or string type of column to a number.', 'start': 1132.427, 'duration': 5.242}, {'end': 1144.235, 'text': 'And the best way to convert yes and no to number is 1 and 0.', 'start': 1138.189, 'duration': 6.046}, {'end': 1146.518, 'text': "So let's see how many columns have yes and no.", 'start': 1144.235, 'duration': 2.283}, {'end': 1148.92, 'text': 'So see all these columns have yes and no.', 'start': 1146.638, 'duration': 2.282}], 'summary': "Replace 'yes' and 'no' with '1' and '0' in text columns for machine learning models.", 'duration': 29.917, 'max_score': 1119.003, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1119003.jpg'}, {'end': 1222.682, 'src': 'heatmap', 'start': 1187.774, 'weight': 0.867, 'content': [{'end': 1199.559, 'text': "OK And so when I execute it, I executed this twice and that's why it is showing me that.", 'start': 1187.774, 'duration': 11.785}, {'end': 1205.864, 'text': "But it actually worked so I'm going to just ignore it.", 'start': 1200.98, 'duration': 4.884}, {'end': 1219.26, 'text': "OK And when you now print the unique values in this data frame you'll see that those yes and no are replaced by 1 and 0.", 'start': 1207.577, 'duration': 11.683}, {'end': 1222.682, 'text': "OK So I'm again going through all the columns and just printing the unique values.", 'start': 1219.26, 'duration': 3.422}], 'summary': 'Executing twice resulted in displaying wrong output, but data was actually correct. unique values in data frame were replaced by 1 and 0.', 'duration': 34.908, 'max_score': 1187.774, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1187774.jpg'}, {'end': 1487.863, 'src': 'embed', 'start': 1424.923, 'weight': 1, 'content': [{'end': 1429.548, 'text': 'for this guy it created three more, and for this guy it created four.', 'start': 1424.923, 'duration': 4.625}, {'end': 1434.212, 'text': 'so out of three we created three and three, six and four, ten columns.', 'start': 1429.548, 'duration': 4.664}, {'end': 1439.189, 'text': 'okay, so you see 27 columns now, Alright.', 'start': 1434.212, 'duration': 4.977}, {'end': 1441.61, 'text': 'So previously there was 20 columns.', 'start': 1440.05, 'duration': 1.56}, {'end': 1444.131, 'text': 'We removed 3 and added 10.', 'start': 1441.65, 'duration': 2.481}, {'end': 1445.452, 'text': "That's why 27 columns.", 'start': 1444.131, 'duration': 1.321}, {'end': 1451.995, 'text': 'Alright So now let me quickly check the data types.', 'start': 1446.252, 'duration': 5.743}, {'end': 1456.032, 'text': 'Okay, so all data types are now numbers.', 'start': 1454.031, 'duration': 2.001}, {'end': 1457.932, 'text': 'There is no string or text.', 'start': 1456.332, 'duration': 1.6}, {'end': 1461.273, 'text': "Okay, so let's move on to the next step.", 'start': 1458.192, 'duration': 3.081}, {'end': 1466.595, 'text': 'In deep learning, the scaling step is very important.', 'start': 1462.333, 'duration': 4.262}, {'end': 1474.935, 'text': 'And if you look at our data, The tenure is in range 6460.', 'start': 1467.595, 'duration': 7.34}, {'end': 1480.499, 'text': 'All these variables are in range 1 and 0, whereas monthly charges will be in some different range.', 'start': 1474.935, 'duration': 5.564}, {'end': 1485.222, 'text': 'Like if you look at monthly charges, see, 1, 00840, et cetera.', 'start': 1480.539, 'duration': 4.683}, {'end': 1487.863, 'text': 'So we need to scale it.', 'start': 1486.623, 'duration': 1.24}], 'summary': 'Data transformation resulted in 27 numerical columns, 20 were replaced with 10, and data types are now all numbers.', 'duration': 62.94, 'max_score': 1424.923, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1424923.jpg'}, {'end': 1542.644, 'src': 'embed', 'start': 1517.737, 'weight': 10, 'content': [{'end': 1524.761, 'text': 'These three columns are not zero and one and hence I want to scale it so that they come into zero and one range.', 'start': 1517.737, 'duration': 7.024}, {'end': 1527.062, 'text': 'And for that we can use min max scalar.', 'start': 1525.101, 'duration': 1.961}, {'end': 1531.876, 'text': 'Okay, so from a scale on, you can import a minmax scalar.', 'start': 1528.433, 'duration': 3.443}, {'end': 1538.281, 'text': 'So minmax scalar will do nothing, it will just convert the values in this range zero to one.', 'start': 1532.036, 'duration': 6.245}, {'end': 1542.644, 'text': 'So if your values in range less than zero to 50, it will convert to zero to one.', 'start': 1538.701, 'duration': 3.943}], 'summary': 'Using minmax scalar to scale values to zero to one range.', 'duration': 24.907, 'max_score': 1517.737, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1517737.jpg'}, {'end': 1617.22, 'src': 'heatmap', 'start': 1557.456, 'weight': 0.709, 'content': [{'end': 1568.41, 'text': 'Okay And when you do this, now I want to just quickly see what happened.', 'start': 1557.456, 'duration': 10.954}, {'end': 1575.177, 'text': "See tenure is now in 0.43, 0.50, so it's in 0 to 1 range.", 'start': 1570.613, 'duration': 4.564}, {'end': 1584.726, 'text': 'Similarly, if you had, I can print the unique values of all the columns by the way.', 'start': 1575.877, 'duration': 8.849}, {'end': 1590.038, 'text': 'And you will see tenure has values 0 to 1 range.', 'start': 1586.192, 'duration': 3.846}, {'end': 1594.184, 'text': 'Similarly, monthly and total charges are in the 0 to 1 range.', 'start': 1590.639, 'duration': 3.545}, {'end': 1601.476, 'text': 'So great, my data frame is ready to be used in my machine learning.', 'start': 1594.906, 'duration': 6.57}, {'end': 1606.933, 'text': 'Now before you create machine learning model, you need to do train and test split.', 'start': 1602.95, 'duration': 3.983}, {'end': 1611.396, 'text': "So let's first figure out what is our x and y.", 'start': 1608.034, 'duration': 3.362}, {'end': 1617.22, 'text': 'Our x is all the columns except churn, and y is churn.', 'start': 1611.396, 'duration': 5.824}], 'summary': 'Data frame is ready for machine learning with values in 0 to 1 range. x is all columns except churn, and y is churn.', 'duration': 59.764, 'max_score': 1557.456, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1557456.jpg'}, {'end': 1652.046, 'src': 'embed', 'start': 1620.102, 'weight': 9, 'content': [{'end': 1622.184, 'text': "And let's do train and test split.", 'start': 1620.102, 'duration': 2.082}, {'end': 1625.126, 'text': 'Again, I have a tutorial on train and test split.', 'start': 1622.744, 'duration': 2.382}, {'end': 1631.451, 'text': "If you don't know what this is, we are just splitting our data set into train and test samples.", 'start': 1625.606, 'duration': 5.845}, {'end': 1639.317, 'text': 'okay. so our train shape, because we are doing 80 to 20% split.', 'start': 1632.111, 'duration': 7.206}, {'end': 1645.661, 'text': 'so 80% of samples we are using for training and 20% sample we are using for testing.', 'start': 1639.317, 'duration': 6.344}, {'end': 1648.363, 'text': 'hahaha, this is 80%.', 'start': 1645.661, 'duration': 2.702}, {'end': 1649.925, 'text': 'this is 20%.', 'start': 1648.363, 'duration': 1.562}, {'end': 1652.046, 'text': 'all right.', 'start': 1649.925, 'duration': 2.121}], 'summary': 'Data split into 80% train and 20% test samples.', 'duration': 31.944, 'max_score': 1620.102, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1620102.jpg'}, {'end': 1763.193, 'src': 'heatmap', 'start': 1708.813, 'weight': 0.85, 'content': [{'end': 1714.519, 'text': 'So how many neurons do you want to have? Well, I will have same number of neurons as columns, so 26.', 'start': 1708.813, 'duration': 5.706}, {'end': 1719.003, 'text': 'OK, so just visualize the neural network.', 'start': 1714.519, 'duration': 4.484}, {'end': 1722.927, 'text': 'Each neuron and the input layer is accepting one feature.', 'start': 1719.403, 'duration': 3.524}, {'end': 1734.531, 'text': 'Alright, what is my input shape? Well, my input shape is 26.', 'start': 1724.208, 'duration': 10.323}, {'end': 1745.942, 'text': 'okay, and this is how you do it, then, By the way, this is not so.', 'start': 1734.531, 'duration': 11.411}, {'end': 1752.106, 'text': 'this is an input layer and this one, the layer we are creating here, is the hidden layer.', 'start': 1745.942, 'duration': 6.164}, {'end': 1752.966, 'text': "It's a second layer.", 'start': 1752.146, 'duration': 0.82}, {'end': 1756.869, 'text': 'So you can have your own like maybe 20 neurons if you want.', 'start': 1753.767, 'duration': 3.102}, {'end': 1763.193, 'text': 'Okay And the activation function, we know the general guideline is for hidden layers.', 'start': 1757.169, 'duration': 6.024}], 'summary': 'Create a neural network with 26 neurons in the input layer and the option to add 20 neurons in the hidden layer.', 'duration': 54.38, 'max_score': 1708.813, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1708813.jpg'}, {'end': 1836.149, 'src': 'heatmap', 'start': 1806.158, 'weight': 0.743, 'content': [{'end': 1807.699, 'text': 'you know what I am thinking.', 'start': 1806.158, 'duration': 1.541}, {'end': 1815.221, 'text': 'you can have this dense layer or you can remove it, because there is input layer and there is one hidden layer and there is one output layer.', 'start': 1807.699, 'duration': 7.522}, {'end': 1825.405, 'text': 'okay, so after you do that, you create model dot compile, where you specify optimize a loss and matrix loss is binary cross entropy,', 'start': 1815.221, 'duration': 10.184}, {'end': 1829.046, 'text': 'because our output is binary 0 or 1.', 'start': 1825.405, 'duration': 3.641}, {'end': 1831.527, 'text': 'Adam is a very commonly used optimizer.', 'start': 1829.046, 'duration': 2.481}, {'end': 1832.548, 'text': 'You can use an Adam.', 'start': 1831.547, 'duration': 1.001}, {'end': 1833.668, 'text': 'You can try different things.', 'start': 1832.688, 'duration': 0.98}, {'end': 1836.149, 'text': 'See machine learning is an art of like experiments.', 'start': 1833.688, 'duration': 2.461}], 'summary': 'Discussing neural network layers, optimization, and binary cross entropy for a binary output in machine learning.', 'duration': 29.991, 'max_score': 1806.158, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1806158.jpg'}, {'end': 1868.334, 'src': 'embed', 'start': 1832.688, 'weight': 11, 'content': [{'end': 1833.668, 'text': 'You can try different things.', 'start': 1832.688, 'duration': 0.98}, {'end': 1836.149, 'text': 'See machine learning is an art of like experiments.', 'start': 1833.688, 'duration': 2.461}, {'end': 1838.43, 'text': 'There is no like golden rule here.', 'start': 1836.69, 'duration': 1.74}, {'end': 1841.432, 'text': 'So try different things, whatever works best for you.', 'start': 1838.83, 'duration': 2.602}, {'end': 1847.674, 'text': 'And then what I always do is I first run like five epochs and just see.', 'start': 1842.692, 'duration': 4.982}, {'end': 1851.196, 'text': 'You know how that is doing.', 'start': 1849.175, 'duration': 2.021}, {'end': 1857.766, 'text': 'So running it for 5 epoch gave me some 80% accuracy.', 'start': 1854.403, 'duration': 3.363}, {'end': 1861.689, 'text': 'So now I have a faith that my parameters are looking good.', 'start': 1857.786, 'duration': 3.903}, {'end': 1868.334, 'text': "If you don't have good looking parameters, meaning you should always try with less epoch and just kind of see.", 'start': 1861.849, 'duration': 6.485}], 'summary': 'Experiment with different epochs; achieved 80% accuracy after 5 epochs.', 'duration': 35.646, 'max_score': 1832.688, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1832688.jpg'}, {'end': 1915.379, 'src': 'embed', 'start': 1888.969, 'weight': 0, 'content': [{'end': 1894.993, 'text': 'Maybe add a few more hidden layers and just kind of see what works.', 'start': 1888.969, 'duration': 6.024}, {'end': 1900.857, 'text': 'And once you run it for 5 or 10 epochs, you will get some feel of where your accuracy is going.', 'start': 1895.374, 'duration': 5.483}, {'end': 1905.08, 'text': 'If your accuracy is increasing, you can increase your epochs.', 'start': 1900.917, 'duration': 4.163}, {'end': 1907.922, 'text': 'So I have a GPU, so it is taking very less time.', 'start': 1905.28, 'duration': 2.642}, {'end': 1910.884, 'text': "But if you don't have a GPU, it might take more time.", 'start': 1908.222, 'duration': 2.662}, {'end': 1914.636, 'text': 'But in the end, I got 82 percent accuracy.', 'start': 1911.345, 'duration': 3.291}, {'end': 1915.379, 'text': 'All right.', 'start': 1914.977, 'duration': 0.402}], 'summary': 'Experiment with hidden layers, monitor accuracy, and consider gpu for faster training; achieved 82% accuracy.', 'duration': 26.41, 'max_score': 1888.969, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1888969.jpg'}], 'start': 869.267, 'title': 'Data preprocessing for machine learning', 'summary': 'Covers analyzing unique values in dataframe columns, data cleaning techniques, one hot encoding resulting in 27 columns, scaling data using min-max scaling, and achieving 82% accuracy in neural network training after 5 epochs.', 'chapters': [{'end': 997.344, 'start': 869.267, 'title': 'Dataframe unique value analysis', 'summary': 'Discusses analyzing unique values in dataframe columns to prepare for label encoding, identifying categorical columns, and creating a function for repeated use.', 'duration': 128.077, 'highlights': ['Identifying unique values in dataframe columns for label encoding', 'Creating a function for analyzing object type columns', 'Utilizing for loop to iterate through dataframe columns']}, {'end': 1457.932, 'start': 1002.989, 'title': 'Data cleaning and one hot encoding', 'summary': "Discusses data cleaning techniques, including replacing categorical values with 'no', converting 'yes' and 'no' to '1' and '0', and performing one hot encoding on three columns, resulting in 27 columns with all data types converted to numbers.", 'duration': 454.943, 'highlights': ['Performing one hot encoding on three columns, resulting in 27 columns with all data types converted to numbers', "Converting 'yes' and 'no' to '1' and '0' for all relevant columns", "Replacing categorical values with 'no' for 'no internet service' and 'no phone service'"]}, {'end': 1652.046, 'start': 1458.192, 'title': 'Scaling data for machine learning', 'summary': 'Discusses the importance of scaling data in deep learning, identifies the columns that need to be scaled, demonstrates the use of min-max scaling to bring the selected columns into the 0 to 1 range, and emphasizes the importance of train and test split by using an 80-20% split.', 'duration': 193.854, 'highlights': ['The chapter emphasizes the importance of scaling data in deep learning and identifies tenure, monthly charges, and total charges as the columns that need to be scaled.', 'It demonstrates the use of min-max scaling to bring the selected columns into the 0 to 1 range.', 'The chapter highlights the importance of train and test split, with an 80-20% split being specifically mentioned.']}, {'end': 1915.379, 'start': 1652.046, 'title': 'Neural network training process', 'summary': 'Discusses creating a neural network with input, hidden, and output layers, using tensorflow libraries, and achieving 82% accuracy after running for 5 epochs.', 'duration': 263.333, 'highlights': ['The chapter explains the process of creating a neural network with input, hidden, and output layers in TensorFlow, achieving 82% accuracy after running for 5 epochs.', 'It details the use of TensorFlow libraries for importing and creating a neural network model with specific layers and neuron numbers.', 'The speaker emphasizes experimenting with different parameters and neural network layers to improve accuracy in the training process.']}], 'duration': 1046.112, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U869267.jpg', 'highlights': ['Achieving 82% accuracy in neural network training after 5 epochs', 'Performing one hot encoding on three columns, resulting in 27 columns with all data types converted to numbers', 'The chapter emphasizes the importance of scaling data in deep learning and identifies tenure, monthly charges, and total charges as the columns that need to be scaled', 'Identifying unique values in dataframe columns for label encoding', 'Creating a function for analyzing object type columns', 'Utilizing for loop to iterate through dataframe columns', "Converting 'yes' and 'no' to '1' and '0' for all relevant columns", "Replacing categorical values with 'no' for 'no internet service' and 'no phone service", 'The chapter explains the process of creating a neural network with input, hidden, and output layers in TensorFlow, achieving 82% accuracy after running for 5 epochs', 'The chapter highlights the importance of train and test split, with an 80-20% split being specifically mentioned', 'It demonstrates the use of min-max scaling to bring the selected columns into the 0 to 1 range', 'The speaker emphasizes experimenting with different parameters and neural network layers to improve accuracy in the training process']}, {'end': 2440.271, 'segs': [{'end': 1969.373, 'src': 'embed', 'start': 1915.459, 'weight': 2, 'content': [{'end': 1916.382, 'text': 'So this looks good.', 'start': 1915.459, 'duration': 0.923}, {'end': 1916.924, 'text': 'Now, let me.', 'start': 1916.442, 'duration': 0.482}, {'end': 1924.497, 'text': 'evaluate the model on XTaste and YTaste, and on YTaste it gave me 80% accuracy as well.', 'start': 1918.133, 'duration': 6.364}, {'end': 1927.399, 'text': 'so this is looking okay.', 'start': 1924.497, 'duration': 2.902}, {'end': 1933.864, 'text': "you know, like reasonable score I'm getting and I am now ready to test.", 'start': 1927.399, 'duration': 6.465}, {'end': 1940.588, 'text': 'so when you do, model.predict on XTaste, YP is Y predicted, by the way.', 'start': 1933.864, 'duration': 6.724}, {'end': 1945.772, 'text': 'so I see that it is returning me the prediction, but the problem is here is two-dimensional array.', 'start': 1940.588, 'duration': 5.184}, {'end': 1959.444, 'text': 'so I want to Convert this two-dimensional array into one dimension and the thing is ytaste that you have is See ytaste that you have.', 'start': 1945.772, 'duration': 13.672}, {'end': 1962.086, 'text': "I'm just comparing ytaste with yp.", 'start': 1959.444, 'duration': 2.642}, {'end': 1967.051, 'text': 'So ytaste is either 0 and 1 and since this was a sigmoid function, this is in range 0 to 1.', 'start': 1962.086, 'duration': 4.965}, {'end': 1969.373, 'text': 'So it could be any value between 0 to 1..', 'start': 1967.051, 'duration': 2.322}], 'summary': 'Model evaluated on xtaste and ytaste, achieving 80% accuracy, ready for testing.', 'duration': 53.914, 'max_score': 1915.459, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1915459.jpg'}, {'end': 2105.381, 'src': 'heatmap', 'start': 2070.945, 'weight': 1, 'content': [{'end': 2078.268, 'text': "I will come back to this, but let's first look into the confusion matrix.", 'start': 2070.945, 'duration': 7.323}, {'end': 2083.331, 'text': 'Now, confusion matrix we have seen in all our previous machine learning videos, so it should be pretty straightforward.', 'start': 2078.308, 'duration': 5.023}, {'end': 2093.155, 'text': "What this is telling you is when the truth is one, which means when in my Excel file, when I'm saying customer is leaving, 225 times it predicted one.", 'start': 2083.851, 'duration': 9.304}, {'end': 2105.381, 'text': 'But 183 times when the truth was 1, it predicted 0, which means this was an error.', 'start': 2097.056, 'duration': 8.325}], 'summary': 'Confusion matrix shows 225 true positives and 183 false negatives.', 'duration': 34.436, 'max_score': 2070.945, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U2070945.jpg'}, {'end': 2258.444, 'src': 'embed', 'start': 2185.196, 'weight': 0, 'content': [{'end': 2186.577, 'text': 'So we have two preseason number.', 'start': 2185.196, 'duration': 1.381}, {'end': 2192.199, 'text': 'OK, so 0th class means 0 means customers who left your business.', 'start': 2186.857, 'duration': 5.342}, {'end': 2201.843, 'text': "So out of the predictions that your model made, which is let's say 889.", 'start': 2192.239, 'duration': 9.604}, {'end': 2204.064, 'text': 'OK, so just a second.', 'start': 2201.843, 'duration': 2.221}, {'end': 2217.631, 'text': 'Okay, let me one second I Let me pull my other notebook because the data is a little bit different here.', 'start': 2207.524, 'duration': 10.107}, {'end': 2220.051, 'text': 'So this is the notebook I have.', 'start': 2218.971, 'duration': 1.08}, {'end': 2222.872, 'text': 'So this is how it looks, 862 and so on.', 'start': 2220.111, 'duration': 2.761}, {'end': 2229.614, 'text': "So here my accuracy was 0.78 and that's what it gave me, 0.78.", 'start': 2223.512, 'duration': 6.102}, {'end': 2236.796, 'text': 'So my precision for the 0th class is the number of correct prediction that you made for 0.', 'start': 2229.614, 'duration': 7.182}, {'end': 2240.091, 'text': 'So how many correct prediction? 862.', 'start': 2236.796, 'duration': 3.295}, {'end': 2249.798, 'text': 'divided by 862 plus 179, meaning how many samples it predicted it to be zero.', 'start': 2240.091, 'duration': 9.707}, {'end': 2251.159, 'text': 'so this is predicted.', 'start': 2249.798, 'duration': 1.361}, {'end': 2253.401, 'text': 'so total 862 and 179.', 'start': 2251.159, 'duration': 2.242}, {'end': 2258.444, 'text': 'it predicted it to be zero and out of that, only 862 were correct.', 'start': 2253.401, 'duration': 5.043}], 'summary': 'Model achieved 78% accuracy with 862 correct predictions for class 0 out of 1041 samples.', 'duration': 73.248, 'max_score': 2185.196, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U2185196.jpg'}, {'end': 2327.79, 'src': 'embed', 'start': 2293.481, 'weight': 4, 'content': [{'end': 2297.284, 'text': 'So 137 sample it predicted it to be 1 but actually they were error.', 'start': 2293.481, 'duration': 3.803}, {'end': 2298.725, 'text': 'In reality they were 0.', 'start': 2297.644, 'duration': 1.081}, {'end': 2300.045, 'text': 'The truth was 0.', 'start': 2298.725, 'duration': 1.32}, {'end': 2301.586, 'text': 'But totally it predicted.', 'start': 2300.045, 'duration': 1.541}, {'end': 2302.447, 'text': 'So 229 divided by 229 plus 137.', 'start': 2301.626, 'duration': 0.821}, {'end': 2304.328, 'text': "That's what I do here and I get 0.63 which is 0.63 here.", 'start': 2302.447, 'duration': 1.881}, {'end': 2305.609, 'text': 'Okay Recall.', 'start': 2304.348, 'duration': 1.261}, {'end': 2314.479, 'text': 'Why the recall for 0th class is 0.86?', 'start': 2312.694, 'duration': 1.785}, {'end': 2322.269, 'text': 'So recall is 862 divided by 862 plus 137.', 'start': 2314.479, 'duration': 7.79}, {'end': 2327.79, 'text': 'okay, so recall is your total truth.', 'start': 2322.269, 'duration': 5.521}], 'summary': 'Out of 229 samples, 137 were predicted as 1 but were actually 0, resulting in a recall of 0.63; the recall for the 0th class is 0.86.', 'duration': 34.309, 'max_score': 2293.481, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U2293481.jpg'}], 'start': 1915.459, 'title': 'Model evaluation and performance analysis', 'summary': 'Covers model evaluation on xtaste and ytaste with 80% accuracy, and performance analysis of a machine learning model predicting customer churn with 78% accuracy, and precision of 0.83 for leaving customers and 0.63 for staying customers, highlighting the importance of independent practice.', 'chapters': [{'end': 2032.878, 'start': 1915.459, 'title': 'Model evaluation and prediction conversion', 'summary': 'Discusses the evaluation of a model on xtaste and ytaste, achieving 80% accuracy, followed by the conversion of a two-dimensional prediction array into one dimension, with a comparison of predicted and actual values.', 'duration': 117.419, 'highlights': ["The model evaluation on YTaste yielded 80% accuracy, indicating a reasonable score for the model's performance.", 'The process of converting the two-dimensional prediction array into one dimension involves mapping the values to either 0 or 1 based on a threshold of 0.5, enabling comparison with the actual YTaste values.']}, {'end': 2440.271, 'start': 2032.878, 'title': 'Model performance analysis', 'summary': 'Discusses the evaluation of a machine learning model predicting customer churn, with an accuracy of 78% and precision of 0.83 for customers who left the business, and 0.63 for customers who stayed, while also emphasizing the importance of working on exercises independently.', 'duration': 407.393, 'highlights': ['The model achieved an accuracy of 78% in predicting customer churn.', 'The precision for customers who left the business was 0.83.', 'The precision for customers who stayed was 0.63.', 'The recall for customers who left the business was 0.86.', 'The recall for customers who stayed was 0.63.']}], 'duration': 524.812, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/MSBY28IJ47U/pics/MSBY28IJ47U1915459.jpg', 'highlights': ['The recall for customers who left the business was 0.86.', 'The precision for customers who left the business was 0.83.', "The model evaluation on YTaste yielded 80% accuracy, indicating a reasonable score for the model's performance.", 'The model achieved an accuracy of 78% in predicting customer churn.', 'The recall for customers who stayed was 0.63.', 'The precision for customers who stayed was 0.63.', 'The process of converting the two-dimensional prediction array into one dimension involves mapping the values to either 0 or 1 based on a threshold of 0.5, enabling comparison with the actual YTaste values.']}], 'highlights': ['Neural networks predict customer churn, emphasizing customer retention (Chapter 1)', 'Telecom Customer Churn Dataset used for hands-on coding and exercises (Chapter 1)', 'Model evaluation with 78% accuracy and precision of 0.83 for leaving customers and 0.63 for staying customers (Summary)', "Converting 'total charges' to numeric format is crucial for accurate analysis (Chapter 2)", "Dropping 11 rows out of 7,000 in a pandas dataframe removes all blank values in 'total charges' (Chapter 2)", 'Over 1000 customers have a tenure of 70 months or more and are not staying (Chapter 3)', 'Visualization provides quick insights into customer behavior based on tenure and monthly charges (Chapter 3)', 'Achieving 82% accuracy in neural network training after 5 epochs (Chapter 4)', 'Performing one hot encoding on three columns, resulting in 27 columns with all data types converted to numbers (Chapter 4)', 'Recall for customers who left the business was 0.86, precision was 0.83 (Chapter 5)', 'Model evaluation on YTaste yielded 80% accuracy, indicating reasonable model performance (Chapter 5)']}