Machine Learning for Everybody - Full Course 

I have no idea how my RU-vid algorithm brought me here while I was sleeping but it made for some strange dreams

I've been trying to learn ML for quite awhile but could never really grasp the algorithim. She explains how the formula comes about and why is it used in the classification or regression so well. My god. Thumbs up for sensei Kylie and free code camp!!!

For anyone getting an error related to converting a list to a float, the model.evaluate is actually returning a list. She has the correction in the code at around 2:05:51, but she doesn't explicitly mention the correction. You just grab the first value in the list (which is why she puts [0]). So change the line where you obtain the val_loss to: val_loss = model.evaluate(X_valid, y_valid)[0]

Thanks for an amazingly simplified approach to ML 👍

⌨ (0:00:00) Intro ⌨ (0:00:58) Data/Colab Intro ⌨ (0:08:45) Intro to Machine Learning ⌨ (0:12:26) Features ⌨ (0:17:23) Classification/Regression ⌨ (0:19:57) Training Model ⌨ (0:30:57) Preparing Data ⌨ (0:44:43) K-Nearest Neighbors ⌨ (0:52:42) KNN Implementation ⌨ (1:08:43) Naive Bayes ⌨ (1:17:30) Naive Bayes Implementation ⌨ (1:19:22) Logistic Regression ⌨ (1:27:56) Log Regression Implementation ⌨ (1:29:13) Support Vector Machine ⌨ (1:37:54) SVM Implementation ⌨ (1:39:44) Neural Networks ⌨ (1:47:57) Tensorflow ⌨ (1:49:50) Classification NN using Tensorflow ⌨ (2:10:12) Linear Regression ⌨ (2:34:54) Lin Regression Implementation ⌨ (2:57:44) Lin Regression using a Neuron ⌨ (3:00:15) Regression NN using Tensorflow ⌨ (3:13:13) K-Means Clustering ⌨ (3:23:46) Principal Component Analysis ⌨ (3:33:54) K-Means and PCA Implementations

Kylie Ying is a gift to humanity

Yesterday I click on a video called 'learning phyton for Beginners'. Today youtube's algorithm sent this video. I was so confuse but somehow listen to it and when I feel I understand something from this explanation, it makes me excited. A genius can make someone understand complicated things, I am very grateful.

*NichesPanel likes this xD* we all know that they isn't, but do you think models buy followers to appear on the internet?

You are literally the best, I've been looking for a tutorial for three days and yours works

Absolutely brilliant. As mentioned in the intro Kylie is a true genius. god bless her

this is perfect! By far the best I´ve found out there, such a clear and complete explanation. Great teacher.

Kylie is such a great teacher and obviously not only understands but applies these topics in the real world. What a great combination, thanks for the course!

I cannot even begin to describe how useful this video was for me. Endless thanks for the priceless knowledge Kylie and freeCodeCamp provided. This video seriously gave me an edge when it came to machine learning. It organized all the abstract concepts I have been hearing about while also describing them perfectly. I have learned SO much. Thank you, thank you, thank you!

I have to agree with those calling this tutorial too hard. I am a professional developer studying Cyber-Sec at the Master's level and found the first hour of the tutorial to be so intimidating that I had to go and learn Python again, just to boost my confidence. I followed it by getting a tutorial on Pandas as well as Numpy, those helped. I came back and realized that, while this is a really good tutorial, it isn't beginner-friendly at all. The kind of stuff Kylie accomplishes in a single line needs multiple lines from me and many more minutes to understand what's going on. As advice to all the newbies, don't be intimidated, try taking the Python basics, Pandas and Numpy courses before attempting this tutorial, perhaps watch the first hour to see what's required and come back.

00:06 Kylie Ying teaches machine learning for absolute beginners 03:29 Importing and labeling data with pandas in machine learning. 10:47 Unsupervised learning uses unlabeled data to find patterns. 14:19 One hot encoding is used to represent categorical data as binary values. 21:04 Data set structure and model training process. 24:22 Training and validation sets are used to assess and tune machine learning models 31:10 Plotting histograms for different features based on class 34:24 Creating train, validation, and test data sets 41:23 Using random oversampling to rebalance data sets 44:57 Introduction to creating a scatter plot to represent data visually. 52:39 Implementing K Nearest Neighbors with SK learn package 55:42 Understanding precision, recall, f1 score and accuracy in machine learning 1:02:36 Bayes rule helps calculate probability based on given conditions. 1:05:51 Explaining Bayes' theorem and applying it to classification 1:13:06 Using probability to make classification predictions 1:16:39 Using MAP and Naive Bayes for classification. 1:23:49 Introduction to the sigmoid function and logistic regression 1:27:53 Logistic regression with L2 penalty 1:35:13 Introduction to Kernel Trick in SVM 1:38:37 Neural networks use neurons to process input and determine output. 1:45:12 Back propagation adjusts weights based on loss. 1:48:19 TensorFlow simplifies defining and training ML models 1:54:54 Training and evaluating a machine learning model 1:58:27 Setting up and compiling the neural network model for training. 2:06:19 Analyzing the accuracy and validation of the machine learning model. 2:09:24 Neural nets and SVM performed similarly, demonstrating different model options. 2:16:19 Simple and multiple linear regression explained 2:20:24 Residual plot shows distribution around line of best fit. 2:27:40 Understanding root mean squared error and coefficient of determination 2:31:32 Understanding total sum of squares and its significance 2:38:07 Creating and manipulating a data frame in Python for data analysis 2:41:29 Plotting and analyzing the relationship between various weather parameters and bike count. 2:48:48 Training and evaluation of a linear regression model 2:52:58 Using TensorFlow for regression with neural networks 3:00:14 Using a neural net with more than one node for prediction 3:04:00 Using neural net model for non-linear prediction. 3:12:12 Discussing the concept of unsupervised learning and introducing k means clustering 3:15:41 Assigning points to the closest centroid 3:23:32 Unsupervised learning techniques: clustering and dimensionality reduction 3:27:10 Principal Component Analysis (PCA) maps points onto one-dimensional space 3:34:09 Predicting wheat varieties based on wheat kernel features. 3:37:41 Unsupervised learning uses no class information for analysis. 3:45:20 K-means clusters and PCA explained 3:48:46 PCA reduces dimensions to 2 for better visualization

falling asleep lands me in odd places

my 7th day - still not finished. Just so nice to see someone do ML work live! Thank you

I just finished taking this course. I took a while. but she explains the theory and examples. I loved the math part of it. she goes on explaining the Supervised vs Unsupervised ML tools and methods. I learnt a lot also with Pyrhon. Thanks for your effort.

This is my first Course which I've completed from FCC, got a good understanding on ML now, Thank you !!

Thanks for the explanations! it is really detailed, her tone is comfortable, I can easily to understand what she said and she elaborates each steps (it is so important as every self learner can know the rationale behind each steps). Hope I can watch another lesson videos of this editor!

Her voice and way of teaching is so soothing. I fell asleep listening to her and I am gonna watch this every night.

This is amazing. I'm a data analyst and had some formal training in machine learning, but my classes were really surface and "teach yourself" style. This is so much better. I also find it easier to listen to women, so that's a bonus lol

Absolutely amazing video, I have watched a lot of machine learning tutorials on youtube and this is by far the best one

Kylie, thanks so much for your time. It was a pleasure learning from you.

Thank you so much for your brilliant tutorials and courses Kylie (please do more!!!)! Could you please recommend some books on the mathematics of machine learning (and books that you found useful when you dived into the subject).

I'm very excited about this. Her neural network video was amazing !

Tysm for covering sooo much so quickly and it was all clear and to the point and I cant appreciate it enoughhhhhh!!!!!

🎯Course outline for quick navigation: [00:00-06:57]1. Machine learning and data analysis -[00:00-00:28]Kylie ying, a physicist and engineer, will teach machine learning to beginners. -[01:05-02:12]Uci ml repository offers magic gamma telescope dataset for predicting particle type using camera patterns. -[05:04-05:38]Using 'fdist' and 'class' to label columns in a csv file to a pandas data frame. -[05:59-06:28]Converting gs to 0s and hs to 1s in data frame class.unique. [07:01-34:39]2. Machine learning fundamentals -[08:00-08:29]In supervised learning, 10 features are used to predict the class label g. -[09:53-10:27]Supervised learning uses labeled inputs to train models and learn outputs. -[13:14-13:45]Two categories: qualitative and categorical data. -[23:57-24:24]Data set is divided into training, validation, and testing sets, with distribution like 60%, 20%, and 20% or 80%, 10%, and 10% depending on statistics. [34:39-44:02]3. Data scaling, splitting, and oversampling for machine learning -[35:13-35:43]Splitting data into train, valid, and test sets, shuffling with 60% split -[37:38-38:10]Imported standard scaler from sklearn for data scaling. -[39:06-39:36]Using numpy to reshape 1d vector y into 2d object for stacking x and y. -[40:28-40:59]Around 7,000 gammas and 4,000 hadrons, oversampling needed. -[42:05-42:37]Oversample the smaller class to match the larger class in the dataset. -[43:43-44:12]Validation and test sets are not oversampled to assess model performance on unlabeled data. [44:02-57:30]4. K nearest neighbors model and its implementation -[44:35-45:03]Introducing knn model for predicting family size based on income. -[49:02-49:34]Using k-nearest neighbors algorithm with a k of 3 or 5 to determine labels for data points. -[52:37-53:10]Introduction to k-nearest neighbors and using sklearn for implementation. -[56:39-57:12]Precision score: 77-84%, recall: 68-89% [57:31-01:36:19]5. Probabilistic classification models -[57:31-58:09]Analyzing unbalanced test data, improving f1 score to 0.87, accuracy at 82% -[01:01:35-01:02:06]Probability of having covid, given a positive test, is 96.4%. -[01:06:00-01:06:32]Probability of positive test given disease is 0.99, probability of having disease is 0.1. -[01:12:22-01:12:52]Naive bayes assumes independence in calculating joint probability. -[01:16:45-01:17:15]Using the training set, we apply the map principle to maximize expression for hypothesis selection. -[01:26:27-01:27:43]Logistic regression fits data to sigmoid function to build a model with multiple features. -[01:28:56-01:29:33]Model achieves 65% precision, 71% recall, and 77% accuracy, outperforming naive bayes but not knn. -[01:33:30-01:33:58]Goal: maximize margins in svms for best class separation. [01:36:21-01:46:45]6. Svms and neural networks -[01:37:13-01:37:41]Exploring the definition and power of svms through the kernel trick. -[01:38:23-01:39:17]Achieved 87% accuracy with svm model, class one scores 0.9, covered four classification models: svm, logistic regression, naive bayes, and knn. -[01:39:44-01:40:12]Neural networks consist of input, hidden, and output layers, each containing neurons. -[01:42:50-01:43:22]Introduction of sigmoid, tanh, and relu activation functions to prevent collapse of the model into a linear one. -[01:43:42-01:44:16]During training, the model adjusts based on l2 loss function to reduce error. -[01:45:53-01:46:26]Updating weight w0 with a new value using a factor alpha. [01:46:45-02:03:23]7. Neural network training -[01:46:45-01:47:13]Explaining the use of learning rate and negative gradient in adjusting neural net convergence. -[01:47:39-01:48:13]Iteratively adjust weights in neural network for machine learning. -[01:55:58-01:56:24]Validation accuracy improved from 0.77 to around 0.81, and loss is decreasing, indicating positive progress. -[01:56:50-01:57:27]Discussing grid search for hyperparameters with variations like 64 nodes and 16 nodes, and adjusting learning rate and epochs. -[02:00:57-02:01:36]Analyzing model performance with loss and accuracy plots. [02:03:24-02:36:19]8. Model training, evaluation, and regression in ml -[02:07:02-02:07:37]Best performance achieved with 64 nodes, 0.2 dropout, 0.001 learning rate, and batch size of 64. -[02:16:09-02:16:46]Using linear regression to minimize error and make predictions for data points. -[02:17:19-02:17:51]Kylie discusses assumptions, including linearity in data analysis. -[02:28:04-02:28:32]The root mean squared error allows expressing error in dollars and same unit. -[02:33:09-02:33:32]High r squared indicates good prediction, adjusted r squared accounts for terms. [02:36:20-03:12:03]9. Data analysis and regression modeling -[02:42:59-02:43:25]Data frame modified: wind, visibility, and functional dropped, leaving 6 columns. -[02:48:10-02:48:42]Demonstrating simple linear regression using temperature data. -[02:54:17-02:55:15]Improved r squared from 0.4 to 0.52 indicates progress in regression analysis using tensorflow. -[03:11:18-03:11:45]Neural net underestimates for larger values, linear regressor also used. [03:12:04-03:22:13]10. Unsupervised learning: k-means clustering -[03:12:04-03:12:32]Linear regressor is limited in capturing non-linear relationships; suggesting the need for alternative models in certain cases. -[03:12:58-03:13:29]Unsupervised learning: using k-means clustering to compute k clusters from unlabeled data. -[03:17:35-03:18:02]Data points recalculated to update centroids for groups. -[03:21:54-03:22:33]Using expectation maximization, reaching stable clusters allows stopping iteration. [03:22:15-03:33:52]11. Expectation maximization and principal component analysis -[03:22:15-03:23:20]Using expectation maximization for unsupervised learning to find data patterns and structure. -[03:24:10-03:24:42]Pca reduces multi-dimensional data to one dimension to capture key information. -[03:25:49-03:26:19]Demonstrating distance using pca to find direction in space. -[03:30:55-03:31:23]Minimizing projection residual to find largest variance dimension in pca. [03:33:53-03:53:51]12. Unsupervised learning and dimensionality reduction -[03:33:53-03:34:22]Implementing unsupervised learning on the uci seeds dataset with three types of wheat kernels -[03:35:06-03:35:51]Importing pandas and seedborne for specific class. -[03:45:46-03:46:12]Using k-means, identified 3 classes based on compactness and asymmetry in scatter plot. -[03:47:54-03:48:18]Pca reduces multiple dimensions into a lower dimension number. -[03:51:32-03:53:49]Unsupervised learning algorithm predicts three categories fairly well without labels. -[03:53:11-03:53:37]Algorithm predicts three categories fairly well without labels. demonstrates unsupervised learning. offered by Coursnap

I love it! I must say this is one of the most comprehensive and well structured videos I've watched lately! Big kudos to Kylie!

Such a great video. I’m already 30min and learning a lot. My major is centered around AI and machine learning so I wanted a sneak peek at what’s to come. Thank you!

Hats off.. cannot imagine that it is possible to explain these concepts in such a simple way. Thank you and please keep on posting such learning videos.

Thank you a lot for this video. This is very interesting and informative. Keep posting like those amazing videos, this is awesome.

If you're getting an error about comparing a list to a float. Changing the "least_val_loss" variable to a list with two infinite floats will fix it. Like this: least_val_loss = [float('inf'), float('inf')]

18:10 As soon as you mentioned "Hot Dog or Not Hot Dog", it instantly reminded me of Jian Yang's classification model from the HBO comedy series Silicon Valley. 😂 But this course is very useful and easy to grasp for beginners like us. 👍👍

Fantastic course! Clear explanations make complex concepts accessible. Practical applications are invaluable. Highly recommend for anyone interested in machine learning. Thank you for sharing!

This worked incredibly well! I can finally play it thanks

Thanks . I am an experienced java lead . I always wanted to learn Machine learning

Enjoy these thorough, clear, visual explanations. She makes what we do accessible to beginners and a perfect refresh for seasoned users.

Kylie and this course are f**ing AWESOME!!! So clear explained!! Thank you so much for sharing!

Woow, this course is amazing. It's presented in a such approachable and undestandable way. Thank you for creating this content.

She is so good at this I woke up yesterday and startng writing an entirely custom Vector Matrix AI learning algorithm from scratch . After watching her explain the concepts from the simplest to the more complex. As a software programmer this has been very interesting to see how simple these algorithms really are.That is how you know you're a good teacher when im not even paying attention and somehow it sticks😁

You guys are awesome.... just search for any tech skill on youtube and bang here is your great videos lectures or tutorials presents in results....

I really like the presentation of your video the time tou take to explain and build the concepts is spot on to get from zero knowledge to yes i understand this.

Ml Full course in Visualised way: ru-vid.com/group/PLDP4z80c2BiL40zhQyiITTSVFve_2e-tJ&si=ktBvnZlHjD2kLkNq

Repost from deep in the comments: ⌨ (0:00:00) Intro ⌨ (0:00:58) Data/Colab Intro ⌨ (0:08:45) Intro to Machine Learning ⌨ (0:12:26) Features ⌨ (0:17:23) Classification/Regression ⌨ (0:19:57) Training Model ⌨ (0:30:57) Preparing Data ⌨ (0:44:43) K-Nearest Neighbors ⌨ (0:52:42) KNN Implementation ⌨ (1:08:43) Naive Bayes ⌨ (1:17:30) Naive Bayes Implementation ⌨ (1:19:22) Logistic Regression ⌨ (1:27:56) Log Regression Implementation ⌨ (1:29:13) Support Vector Machine ⌨ (1:37:54) SVM Implementation ⌨ (1:39:44) Neural Networks ⌨ (1:47:57) Tensorflow ⌨ (1:49:50) Classification NN using Tensorflow ⌨ (2:10:12) Linear Regression ⌨ (2:34:54) Lin Regression Implementation ⌨ (2:57:44) Lin Regression using a Neuron ⌨ (3:00:15) Regression NN using Tensorflow ⌨ (3:13:13) K-Means Clustering ⌨ (3:23:46) Principal Component Analysis ⌨ (3:33:54) K-Means and PCA Implementations

Thanks for making the excellent course!

It really worked for me after I look and try some tutorials, yours is the one that worked. Owe you a lot.

The best teacher I have ever seen in this subject...!

This is basically Machine learning for everyone who is a Python programmer. But seems to be a nice video. Will get back to this after I finish Python

Kylie explains better than my professors LOL. Great job FCC and Kylie, thank you very much

Prefect course for newbie. Teacher explains very good. 🎉

A great introduction into machine learning, even for someone who may not be much familiar with Machine Learning, can learn a lot.

Yes, as everyone is the audience let's start with an example that few can relate to and then just jump into code, again, everyone is well versed in code, without any explanation or overview.

Loving this course. Very Intuitive and practical.

41:52 1:02:49 Just reminding myself where I stopped the video. Absolutely fantastic walkthrough!

This was definitely valuable! Thanks for sharing Adnan!!!

It would be very beneficial for beginners to make Shallow, Deep and Convolutional Neural Networks from scratch. Because by doing so, they can learn many activations and their derivatives, forward propagation, and backward propagation. Along with, the dimensions of matrices and what is actually happening at each layer. Later on, they can shift to Tensorflow or Scikit when doing professional learning but I do advise to at least implement a neural network from scratch.

The best introductory machine learning video! I have tremendous respect for Kylie. I think it is the clearest and easiest to understand explanation in the world. Thank you so much.

I think it is a very advanced video honestly. it should mention that you need a very solid foundation of statistics, coding, data science to even understand what is going on!!! its a medium level video

Did a refresher on ML as it's been a while... Kylie rocks! Delivered with the right amount of theory combined with examples and importantly, clarity! Top marks!!!!!

I’ll finish this course this time around hopefully 🙏🏾

@Freecodecamp you're always teaching us great skills with great mentors. Salute sir 🤗. Kindly upload the electric vehicle course if you have a great mentor available.

THANK YOU for making her course available on this channel !!!!!!! 👍😃👍

She is so smart and makes things simple to understand👍

This is a great course and thank you so much. I was able to understand and learn so many new things! Some topics I had struggled a bit were quite clear through this course.

This one is great video regarding to the Data Science and Machine Learning.

The best course ever. THANK YOU!!!!

Super useful video tbh. Once she's started going off with the coding it was amazing . Hope to get to that level one day

Thanks for sharing such an amazing content!!

Great video...stayed up all evening watching; wondering if you would demonstrate ketosis distribution measures or is that too difficult to script out?

thank you Kylie for this great video. I loved some of the workarounds you found to simplify, like when it's binary not needing to use an encoder I didn't know that. I changed my code around 34:00 because it initially confused me the two uses of 'label' in the code so went with "for col in cols". I wasn't sure about your data split, is it not better to randomise selection? Can you be sure that the data is not reported in the dataset in some sequential way?

Ironically, she is so humble as well. She is great. Thanks for making this.

Great course! Thank you for the hard work!

A great introduction into machine learning, even as someone who is into the tech industry, I am not too familiar with machine learning.

Amazing course, I coded it along in VSCode. This made me learn quite a lot about conda, environments, etc. With this kind of hands on learning, not only I believe I can start doing data analysis and creating my own models and understanding their predictions. Thank you all so much! 👏Great work! (And well, it's not for Python noobies... Anyone past the basics with most of the libraries used can understand it)

It would be helpful for people without coding experience to explain the libraries being used, the languages, and how it works. Or to explain the bare minimums required. This seems more like it's geared to people with python programming experience as a base

What have I woke up to ?

thank you FCC and kylie ying for this amazing ML course....best for recap in short time about classical ML domain

This is the best course ever on youtube about machine learning . She know the fundamental and know what to teach . ❤

It seems half of us are here after falling asleep

Thanks for the content. This isn't for everyone though.. She speeds through so much and beginners are just sitting here going "What the hell is going on...". Looks like that was being pointed out early in the comments and then there seems to be a flood "this is so clear and fully fleshed out!" comments that have been pushed to the top.. That's a bit sus.

It's 2:16 am and I just woke up to this on my screen and yes I wanted it so badly. So strange but it's now downloaded and will watch and learn when I wake up. Thank you

Thanks guys for your great effort to educate the world.May God bless you

I was so excited to see this course because usually freecodecamp courses are for beginners but this one is not. You need to have good knowledge of numpy and other libraries to be able to learn more from this course. Really disappointing. They should let users know the basics that are required for this course. But ultimately this is a very good course and it's free so really appreciate it guys. Thanks

Is there a summarised version of this lesson? perhaps for someone that isn't intending to gain a fullon knowledge but gain a sufficient, generalized idea of every subject Kylie is covering :),

@ 39:09- Since X and y start out with the same dimensionality (both 2D, with y as a single-column dataframe), do we still need to reshape y?

Thank you so much for this amazing video! I had to take a nap halfway through because there was so much to digest haha

train,valid,test=np.split(df.sample(frac=1),[int(0.6*len(df)),int(0.8*len(df))]) key error 0??

Hi, thank you very much for this high quality course, a real master piece. Maybe as a newbie, I could be wrong but it seems there is an error in the explaination of the data preprocessing (usage of standardscaler). The standard scaler has to be fit and transform the train sample however, the scaler parameters should then be stored and it has to ONLY transform the next sample (valid and test). Fitting the scaler on the valid and test sample could have a bad effect the the X values and misaligned the data from the train sample. Once again, I'm new on this topic but it seems quite clear this standard scaler usage is inappropriate.

This is a really good course. I would like to see live example predicting from the dataset. I got thru the first 30 minutes with my basic python and Pandas but started to get bogged down afterwards. Still - getting thru this video will give you a practical basis and reviewing it again and agumenting with with further study would be a sensible approach. Right now ML seems a lot harder than I thought it would be by having to fuss and transform the datasets. But keep at it and it will click. Thank you Kylie for this excellent tutorial.

Great course, excellent combination of material with application examples.

Can someone confirm if this goes beyond the basics and actually helps you to build stuff? I’ve done plenty of these courses and most of them only contain basic information about a 100 different things and very little demonstration of actually building a significant project from raw data. So, my fundamentals should be fine after a bit of a brush-up but I need a course that goes beyond.

Hey, thanks for sharing. Great video. I get an error at this code: train, X_train, y_train = scale_dataset(train, oversample = True) ValueError: input contains NaN, infinity or a value too large for dtype ('float64') Does anyone else face this issue /know how to solve it?

Thanks for creating this awesome tutorial for free ❤

Kylie, can you please explain to me why would the two lines at timestamp: 3:02:53, appear to be the same function when I plot them?

Thanks for this great tutorial freecodecamp. We also need statistical data analysis with Python and/or R (especially descriptive statistics, probability distributions and hypothesis tests , anova etc... )

Awesome!!

GREAT Course! Wish we could have more videos from HER!!

Totally awesome! Thank you for Sharing! 💯✴

There's no way you're putting out all of this for free. Better explained than in school ong