Getting started in scikit-learn with the famous iris dataset 

Having problems with the code? I just finished updating the notebooks to use *scikit-learn 0.23* and *Python 3.9* 🎉! You can download the updated notebooks here: github.com/justmarkham/scikit-learn-videos

Your clarity in speech your thorough explanations are absolutely outstanding. I've spent 5 years at university and have recently finished a triple major in math and sciences, and I must say that you are an excellent educator. Your teaching skills easily surpass the majority of professors' I've experienced. Very nice work. I have been accepted into a CompSci PhD program for this upcoming fall semester, and I am extremely interested in machine learning. Consequently, I have been trying to self learn data science in the meantime to better prepare for my studies, and finding good tutorials that offer lucid explanations has been quite difficult. Andrew Ng's course is wonderful, but lacks the Python element, a language I am very keen on and will be using in my graduate studies. Others are difficult to follow, put together in haste, or offer no explanations - just code. Thank you for the effort you've put into creating these videos, and for your incredible teaching abilities. Keep up the fantastic work, and I wish you the best of luck for your future endeavors - although I'm sure your skill is what will enable your success, not luck!

Just want to say you are an amazing speaker/presenter. It's inspiring to see thoughts in a human so collected and clear-headed. I can only aspire to.

Thanks a ton brother !! one of the best tutorials for a python library. the clarity in explanation is 10/10

Can't be more simpler than this. You are an awesome teacher. I've seen many tutors and videos, no-one is so concrete, simple and your explanation style is outstanding. Looking forward for some complex ML topics.

Goodness. These videos are amazingly useful and the best of their kind I've found online. Thanks, Data School, for putting them together!

I never imagined something as archaic as machine learning could be taught with such ease and grace. Everything about your videos is spot on. From your lectures, to the development environment, to the recommendations and links. Love the follow up links. So helpful. Can't wait for more!!!

Your Explaination is so good, referring to 10minutes of this content helps me to know things much more better.

Clear and concise explanations. You've covered things other tutorials I've seen have missed. Thanks!

I never thought that some would teach so clearly. Even a layman would understand this stuff. Great, thumbs up. U mentioned the data have to be in a shape but what if some data are missing and not in a proper shape

I recently found this tutorial series while I'm looking for something to learn about machine learning and I'll just say Thank you! This is especially useful for those people whose mother language isn't English! Thanks again!

I was looking some easy but effective tutorials on ML. I have no doubt, this series will be that what I was looking for. Thank you very much for taking this initiative. (Y)

Thank you so much for the tutorial. I love it! You speak so clearly and concisly! I cannot wait longer for your coming video!

Thank you for explaining this slowly - it makes it very easy to follow.

Im just pausing the video to say that your explanation couldn't be better. So clear! THANK YOU !!!

A smooth and calm voice. It is easy to absorb.

A great video i have ever watched , because it gives clearity and full understanding most importantly also provides resources, heads off to u man, what else one need

Thank you for making some really structured and good content. I have learned to use pandas in python by watching your videos. The way you explain is clear and structured. Please, keep making more videos so that most people like me will be able to learn even from halfway across the world. Keep up the good work!

thank you! In my opinion, the best tutorial on the topic I found on the net. I know Python and I'm approaching machine learning ... very nice

I have seen a lot of stuff on machine learning but couldn't get it. Your work is amazing. You let me know that I can learn ML too. Thank you soooooo much.....

you are the best teacher :) Thanks for these videos

Just Love the way you teach. Point to Point.. Thanks a lot for the series

Thank you so much for these videos. This is exactly what I needed to get started with Machine Learning

Thanks so much for the great video, and specially for introducing the , now I finally know why should we use Numpy! And seems very simple to use!

Hands down the best machine learning tuts on youtube

Excellent classes!! Amazing for those who have english as a secound language like me!! Tks a lot!

Quick tip for new users to Python like me: in Python 3 you need to wrap the arguments to your print statement/function in parenthesis. So, print iris.data won't work but print(iris.data) will.

Absolutely brilliant and clutter less. Thank you

Excellent - well thought out and lucidly explained.

This is "REALLY AWESOME".....Great work...Thanks a lot for these awesome tutorials..

Very helpful videos- clear and precise. Thanks!

*Note:* This video was recorded using Python 2.7 and scikit-learn 0.16. Recently, I updated the code to use Python 3.6 and scikit-learn 0.19.1. You can download the updated code here: github.com/justmarkham/scikit-learn-videos

thank you: you speak very clearly an dslowly which is perfect for a non english native. And course is very clear and step by step: perfect

Awesome stuff man! very clean explanation and easy to learn. cheers

Very useful and easy to understand. Thank you

These videos are so useful. Thanks.

You are a great professor! I hope all the best success to you ;)

thank you very much to post the videos, you are my role model :)

Thank you. . You are gifted , you made it simple

your voice is amazing and explanation is also very good....you just seem like Sheldon while explaining

This video helped me a lot. Thanks.

Dude was doing epic shit in 2015, salute :)

Very good explanation.. Liked it (Y) .. Thanks for other links as well..

I Love your videos! Great Job and good luck!

Fantastic video series :)

Still very relevant there is just newer versions of the same thing (IPython is now Jupyter Notebooks and version numbers of the same library were expanded and improved ) code that will still work for a demo. Fundamentals don't change anyway and the delivery of the tutorial remains excellent to this day. I like to access data locally from my hard drive needing no internet to read IRIS data as HTML location but data frame for Pandas instead - it wouldn't hurt to read the text file into excel and convert that to .csv file instead.

This helped alot, thanks!

Thanks a lot, you are a perfect teacher.

Awesome videos!!! Thank you so much

in this video you said that we know the target as 0,1,2 so we know it is a classification problem but not a regression problem. But according to my understanding, we can apply multi-class logistic regression in this problem.

Very well explained

YOUR VIDEOS are awesome

Amazing lectures :-) You remind me of sheldon cooper too

Thank you very much

Thanks for these videos, these are a great help for getting started! Quick question, I may have missed it since I skipped a couple videos since I already had anaconda and Jupyter notebook installed when starting, but are you using Python 2.7? I am using 3.4 and as I am going through I ma getting some syntax errors and such. Nothing that's too difficult to troubleshoot or anything, I was just wanting to know if I need to keep that in mind throughout the series?

A great resource for machine learning .....

Awesome explanation

very nice explanation

Awesome video

watch this on 1.25 , thats much better, thank you sir

Hi, Kevin, thank you for all great videos. Is there any easier way to import sklearn modules? Like from sklearn import all? Thank you!

Excellent

Thank you!

I have a little question: why do we have to predict data based on informations that we already have? e.g. when i defined the X matrix to the features and the y vector to the predicted data, all we want to do s to verify how accurate the computer is to analyse, learn and predict which group does it belong to?

Your voice is amazing.

Love all your videos.could you make videos on coding machine learning alogrithms?

Omg. This is excellent. Will you be coming up with tensorflow tutorials?

Great tutorial! I have a question if I had a dataset that is not a toy dataset -not built in sklearn-, how do i import it? what function do I use?

yes, very amazing voice as well.

Great tutorial! I am sure this series will become an excellent resource for all machine learning enthusiasts. Bundle of thanks! I have a question: What if there are more than 1 observations (rows) associated with an outcome? For example, we want to predict if a certain vehicle is a car or truck using labelled data and each vehicle has, say, 10 observations in a second? Would we consider all of them in the X matrix?

Thank you for your videos! Can you tell me, is .target_names an action, and one that can be used on all scikitlearn datasets in Python? Or is it specific to the iris dataset? Thanks again!

Good explanation. You need to keep it up the good work. Sadly though, we rarely can find methods on how to create dataset exactly like iris or similar rather than explaining from finished data onwards. Mostly, it would be at the extreme ends i.e. either show how to create data set or from a ready dataset to machine learning which hopefully won't deteriorate the passion of beginners to improve their skills.

I am a beginner in machine learning. The concepts are well-explained, thank you! Btw, I watched the video with 1.5x speed. haha

Hi, Thanks for the video series. They are really helpful. In the video above you have described that scikit learn always takes numeric values, we need to convert the categorical to numeric. But how would it segregate between which columns have numeric values & which other columns are categorical. Please share some links to read more about this topic if you have some

BTW, I really enjoy this speed for education purposes. Rarely do I get an opportunity to actually digest each sentence properly.

Hi Kevin, how can I feed text data into ML model. Like what pre processing steps I need to perform over that data so that our model will be able to understand it. Or how we deal with such kind of data to train our model?

amazing =)

I love your video series. Can you make a video teaching the above using .csv file and not database. Also, while predicting accuracy of model made, is there any method to make train data and test data? If it's not mentioned which column is my target data then what is the approach to determine the accuracy? Please make a lecture on this.

I have a question, when we do a normal train test split on a dataset our x and y are pandas series objects and not numpy arrays, but they still work,pls elaborate on this point

Great video. your teaching style is awesome and also explanations. I am new in machine learning so i fell some difficulties to understand.. please give your advice to improve my understanding.

As every week only a single video will be published, can you please share some other resources as study materials? It might be any video/book or whatever you like to share. Thanks again!

Hi i really am enjoying this video series so thank you for that. I am just learning python at the moment do you think its harder to get a job in machine learning compared to say django? I find machine learning more interesting but I feel its harder and there are fewer jobs in the market. Im from UK btw, thanks again.

Thanks for the material. If you are following ISLR book, could you share the chapters that we might read along the way, just like how you added to the first video.

Wow, I think x,y are like in math in high-school when we set an independent variable X (which is the observation right now) and we were getting a result y(the target).

Hi, Very nice video and well explained. As mentioned sklearn only uses numeric for features and response. How can we convert values into numeric for classification or regression? I know we can manually assign 1,2,3 .... for each value but does sklearn has a library which can convert categorical to numerical.

Can you also do a tutorial on how to create your own dataset, that would be amazing :D

Thanks!! This video was exactly what I was looking for. You mentioned that you will explain how to import our own dataset. Can you please point me to that video?

I want you to tell me what did you do for your vocal it's natural ,it's so cool

1. Hi, at 08:10, you said that predicting wether an email is spam or ham is also an example of classification problem. I dont understand how is that, Can you please elaborate that. 2.One other question I have is that if in my experiment, target can be coded either by two features, one numerical discrete and one numerical continuous (it can be coded by both individually, its upto me which feature to use to code my target). Then will it be regression case or classification ? I am asking this bec you said that if response being predicted is ordered and continuous then its regression case. (so if i chose the continuous feature (which is rounded to nearast decimal), then can it be a regression case)

Thanks for the tutorial. However, I'm interested in using data from my own excel table (so CSV file) rather than a pre-compiled dataset. Is there a way to convert this to a Bunch form that scikit requires?

Will classification problems only deal with mutually exclusive sets?

You remind me of Sheldon Cooper. Nice video btw.

Are you also going to create few on unsupervised n dimension reduction concepts

is respone and target the same?

i am not able to use scikit-learn module as skllearn. Stating an error sayin : no module named "skllearn" found.

Great tutorial. Can I ask how would you set feature_names and target_names on a dataset? sorry if that is a stupid question!! I was trying data.feature_names in a dataset downloaded from AdventureWorks2012. and giving me an error as below: AttributeError: 'DataFrame' object has no attribute 'feature_names'

In playing with scikitlearn I found that it doesn't much mind strings as targets. To test this I created a simple classifier of colors where I gave it the HTML codes of a few basic colors, then by feeding it codes of other colors it would classify , for example, pink as a type of red. Or it would classify aqua as a type of blue. In doing this I gave it a list of color names as the target. During prediction it returns the string within an array marked with a byte-type. In principal why is this a bad practice? Thank you for your videos.

How can I make my own dataset in sci-kit learn? Do I have to make an excel file first and then download it as a csv

thanks for amazing and well-explained tutorials. I have a dataset of depth information from Kinect and they are in .bin extension? How can I load them and massage?

Hi, are there videos about how to prepare the raw data before building a model, like how can i turn my target whose type is category to number? And I want to ask how can the model tell the target number represents a category instead of numeric so there is no scale or distance between two target number.