Machine Learning: Testing and Error Metrics

[@victorrabvukwa79]

Your online tutorials are really great. You demonstrate intimate understanding of the subject and deliver it so so well. It's just left for me to take it in. Congratulations on this series. Immense help to students.

[@andyn6053]

Your videos are without any doubt the most easy to follow and easy to understand out there! Thanks for explaning things in a simple way so it finally makes sense!

[@WeiranYe]

Thanks Luis! This is the most concise introduction to those terminologies one could find on the web! Great job. You saved me tons of time! Thanks a lot!

[@SerranoAcademy]

Thank you, Weiran, glad you liked it!

[@prempant6428]

@@SerranoAcademy Hey, could you please explain the difference between in-sample risk and training error? And how we can measure the in sample risk thus, we can compute the optimism of the dataset?

[@waseemanwar3327]

Thanks Luis. This is the simplest and best ever tutorial in ML I have come across.

[@robertknight9242]

Incredible video! I always find remembering evaluation metrics difficult but this is a really great way to get my head around it in a memorable way! Will definitely be watching all the videos you put out - thanks in advance.

[@vmbijli]

One of the best explanations on ML. Amazing clarity of thought. You have a super visual mind and thank you Luis for sharing what you see with us. Ever grateful 🙏

[@mshans66]

Awesome! What a way to simplify the complex concepts! You validated once again that you are awesome and the hyperparameter used is your pedagogic style!

[@franciscovinueza5320]

You are the best. You explain better than my Machine Learning Lecturer. The use of images, colors and explanation are 10/10.

[@abdulqureshi208]

Your Videos are one of the best on the web, please keep it coming.. Thanks

[@zirakqader3664]

I would thank you first. The way that you are simplifying and giving examples is really great. Well done :)

[@AlvinRyellPrada]

Thank you for pointing me here. Done watching the entire video and despite that English is not my primary language, I am hooked and fascinated how you make the explanation exciting! The visuals and your story telling is absolutely superb as well! Your channel is a God send!

[@randyluong6275]

concise, vivid. one of the greatest tutorials I have come across.

[@blesucation4417]

Just want to leave a comment so that more people could learn from your amazing videos! Many thanks for the wonderful and fun creation!!!

[@shaiknazeeruddin2901]

Love the way you teach and make complex things so simple. Thanks a lot Luis. Hoping to see more videos such.

[@donsweboshui8099]

Awesome Luis, it is the simplest and easiest tutorial for ML I have ever see from KZbin, thanks again. Keep it up.

[@uchennaudeh9816]

Thanks Luis. Exceptional delivery. You really have an innate understanding of these concepts and algorithms. Thanks a lot

[@trinadhsingaladevi8918]

Hello Luis, First of all your videos are great. Thanks a lot. In the video, the precision for credit card fraud detection is 0/0 right? how come it is 100%? Am I missing something here?

[@syedali9198]

Haven't watched the video, but I am pretty sure it would be amazing. You are doing amazing service to so many people. Thanks.

[@virtualtinker2303]

Video should get way more likes and views, very detailed but in a simplfied way. Thanks a lot for the great info!

[@2parinda]

Best explanations I have ever seen regarding ML. Thanks a lot for Your effort

[@avinashtandle1489]

Thanks Luis for making, model evaluation attributes simpler

[@houyao2147]

I like the discussion of recall/precision. I was always confused by how to get an intuitive explanation of recall and precsion, now it's clear.

[@anindyachaudhuri9880]

This is extremely good, nicely made and presented. Thanks a lot.

[@georginaarno9880]

Thank you Luis! You and the material you use are really good for teaching. They increase my interest! I alteady bought your book "Grokking Machine Learning" in digital version. Can't wait to read it and work with it!

[@SerranoAcademy]

Thank you so much Georgina! I hope you enjoy it! :)

[@pcarlisle5688]

This is so easy to understand. Just the best tutorial ive seen for this topic!

[@nikhildethe2973]

Hey Luis, I saw your ML videos, and they are awsome. I understood the core and core of it. wow... Thanks maaaan...

[@MyJackzhang]

Thanks for the great video. I really enjoyed all the videos you posted on ML. Hope more to come.

[@SerranoAcademy]

Thank you Chaoli! Just added a CNN video, check it out!

[@firehawk97]

This just answered so many questions I had about ML accuracy scoring and HOW the background code in R-Studio (many built-in functions) are actually calculating accuracy percentages. Thank you for this video

[@speakingmia7298]

Thank you Luis.Machine learning seems more friendly due to your vivid explanation!

[@juanmateu19]

Increíble Luís, no hay nada similar ni en udemy ni en youtube. Felicidades por conseguir se tan eficaz a la hora de transmitir tus conocimientos y mil gracias por brindarnos este contenido. Te deseo muchos éxitos!

[@SerranoAcademy]

Muchas gracias Juan! Que lindo mensaje, me alegra que te guste el contenido. Abrazo!

[@nikhilbharadwaj8972]

Brilliant explanation. Thanks for taking the time to make the video.!!

[@randunik77]

I enjoyed all the videos you posted on ML. Great videos. They make me better understand a lot of concepts and terms that I head/read in my efforts to learn how AI works. What I would realy like to see and understand now, is how these concepts are translated into code (tenser flow or some other framework). Thank you for your videos.

[@alainiliho1019]

I have really appreciated this video. It helped me understanding I was facing overfitting in my project. Would you mind doing another video about regression problems? Thank you so much in advance.

[@adityaverma3095]

Thanks Luis! This very well summarize the model selection in a very concise manner. Can you please do a video on various metrics to assess a machine learning models, like lift chart, ROC curve, confusion matrix etc. all combined and their use in different cases?

[@pavleenkaur6903]

Thanks Luis for this detailed explanation! Request you to upload videos on some other expansive metrics too such as AUC, ROC etc.

[@marcerodriguez7553]

Gracias Luis ; excelente trabajo . Salu2 desde Argentina .

[@MV-qm7rs]

Awesome bite-sized videos Luis, really intuitive to understand! Great job!

[@alonsodiaz287]

Luis ! Great Course ! best one out there . I will recommend the IBM QMS Quality Engineers to view this video! thanks !

[@biswasstar]

Very helpful. Great Explanation!!!!! Thanks a lot :) Would love to have more videos giving clarity of statistics concepts used for machine learning .

[@aminarahman2429]

Your videos are awesome. That's my 3rd watched video today, made by you. This is really awesome since it clears the basics in an interesting and simple way. I wish I had watched it earlier. That'd have helped me in my previous vivas. :P

[@bla-ig4bd]

Love the three golden rules. Rule #4: don't forget the first three

[@saurabhiim]

Hi Luis , Many thanks for these interactive videos & very nice explanation of the complex topics on data science ...You are very clear about the fundas of the subject ...I request you to kindly help us with Random Forest techniques & SVM . Thanks in advance

[@sebascol]

Great video, great explanation!! Thanks Luis!!

[@elgs1980]

Nobody else explained things as clearly as you did.

[@gitadanesh7496]

Thank you very much for the easy and amazing explanation

[@arminehayrapetyan3373]

Why do we get 100% precision score in Credit Card Fraudulent problem? Is that correct?

[@balajee41]

Wow..I don't think there is a better explanation than this

[@jenniferaduwo6635]

Wooh great and detailed explanation thank you, it is easy to follow, I would love to know your thoughts on when to use AUC metrics for model testing

[@MrJeevan415]

Thanks Luis, for explaining in a subtle and easily understandable way. But can you please share the link of book (Grokking machine learning that you. wrote) As it shows there in publication and estimated is July2021 (which is also not guaranteed). Any faster way to get the book.?

[@celismaroliveira6081]

Thank you so much! Another amazing tutorial!!!

[@ruudhermans4243]

Great video! Going to check out the rest of them.

[@petesantago5977]

For k-fold. What model do you use? Since you trained on 4 different training sets, you have four models.

[@SRAVANAM_KEERTHANAM_SMARANAM]

Sir, What software do you use to show the points as blue and Red points. The presentation looks beautiful with colors.

[@cahitdemir2756]

Thanks a lot, you made very easy to understand metrics

[@vamsikrishna1131]

Awesome examples & explanation. TY

[@92hinga]

Thank you Luis. Well explained

[@PS-kn5lr]

This is amazing!!! Thank you sir...I am trying to learn ML since a long time...after so many complicated tutorials I lost interest in ML and went in web development . Now I found this tutorial and could grasp all those complicated terms with ease. I went to your Udacity ML nano degree course link but its expensive :(

[@SerranoAcademy]

Hi Parigha! Thank you for your kind message. Check out this free deep course that I teach with other people Udacity: www.udacity.com/course/deep-learning-pytorch--ud188

[@ujjwaljain6416]

Underfitting -> Over simplifying things -> Trying to kill dinosaur with trap -> Error due to High bias -> bad on the training set -> bad on testing set Overfitting -> over Complicating things -> Trying to kill bee with bazooka -> Error due to High Variance -> Great on the testing set -> bad on testing set I did struggle with this for a long time

[@scottlebowitz9258]

sorry if this was asked or explained in video, what is cross validation, how did it take the quantity of the training set and the qualities of the testing data (seem so from the graphical depiction? And if it did take data from the testing set, isn't that still a conflict to the rule of not using testing data? Thanks, great videos, may take some classes online.

[@vinaysannaiah2817]

cross validation is a data which will be kept separately from the beginning in order to measure and tune the model. It is not part of either the training data or the testing data.

[@carvaka100]

Thanks Luis, A very well presented explanation.

[@timetraveller7513]

Thank you so much Luis, grt session 👍🏻

[@Adnan25048]

Great video. Great explanation.

[@lim-chanconnie1244]

Thanks, it is better if you can help to teach us on codes as well for some real Tensorflow and Keras example for image recognition. I like your style. Very clear as I had some basics but yours clarify my basics to better level.

[@seyedmansourbeigi9126]

Luis great another masterpiece "Machine Learning: Testing and Error Metrics" thanks

[@IqweoR]

Thank you very much. Very informative. Going to try udacity courses now!

[@ErikFKL]

Thank you Luis! Really appreciate your works.

[@khaldoonalhussayni1853]

great job, it is so useful and simple thanks so much, but how can I find parameters and hyperparameters for other algorithms such as Naïve bays and K-nn

[@mariogalindoq]

Luis: nice video, but why did you kept hidden the F_beta expression? I think you should show it.

[@FebiyanRachman]

For the credit card example in 21:35, won't the precision be 0/0 since nothing is classified as bad?

[@charlainebrowne8715]

great series made easy to comprehend thank you

[@watchsatsang]

I am very much thankful to you for so nice videos.

[@sunday8983]

This is brilliant! Thank you!!

[@kevinha3192]

Thanks man !!!! This is so very helpful !!!!

[@soajack]

Great lecture Luis ! Thanks !!!

[@gkprasad100]

Awesome explanation !

[@ElVerdaderoAbejorro]

Hi Louis. Thanks for your video. I don't understand why you said we are breaking the Golden rule at the end. We are not using the test set to train anything, just to evaluate the error. In other words, at no given moment the test set influence the parameters of our model. Am I missing something?

[@reyhanehhashempour8522]

Hi, we didn't use the test set to train our models, that's right; but we used it to choose the best model! so we made a decision based on the test set which is against the rule. i hope I answered your question.

[@holgeriwersen5367]

Very good and helpful. Easy to understand in the beginning , at the end a little bit too fast about too much new things .

[@bhaskartripathi]

Are all of these evaluation metrics precision, recall, F1 etc only for classification ? Is there a video for regressors as well ?

[@deenuy]

Wow! Brilliant session

[@thisismrsanjay]

so clean explanation

[@sibusisokhanyile8181]

Luis you are the best man!!!!

[@jakovbazsanov1725]

Awesome ML training! Thank you, Luis!

[@romeshdoshi3161]

Very easily explained

[@explorewithpardeep5442]

TBH this is literally an amazing explanation of the problem i get till to 3 years

[@sreejeshmohanan5155]

I Dont think there is a more PRECISE explanation for evaluation metrics than this, in Internet

[@mahimachandane9067]

Great explanation 🙏

[@amralaa1834]

Great explanation!, can anyone tell me what's the program that's used for making this presentation? Thanks.

[@MyStudyIsFun]

Awesome intro on ML model evaluation

[@timb8140]

One of the best explications, please keep it up. Subscribed :)

[@victorrodas4357]

Muchas Gracias, Luis!

[@TheDjomed]

Hello Luis, Thanks for the presentation, can you provide us with the slides in pdf if possible ?

[@savvasemexides6557]

Very helpful video, thank you very much!

[@SiriJustDoIt]

Wow... Simply Super ... F1 is 1

[@eusebiusballentine3187]

Great stuff Luis.

[@ashjanalsulaimani4537]

Great explanation, Thanks

[@rsinh3792]

Sir reviewer has asked me this question I don't know how to address it, can you please guide me "Use some statistical significant test such as T-test or ANOVA to prove you validate the proposed diagnostic model on patients and quality improvements of your method". I have two datasets. Dataset 1 was used to train the model and dataset 2 was used to validate the trained model. I have trained the ML model deployed it and Validated it on new data and presented the results. Actually, I have understood the question. Shall I apply the statistical test between the performance metrics of trained model results and validation results? Please help me, sir.

[@leilayousefi1512]

Thank you for the amazing video. Could you please share the slides?

[@relevelschool]

Thanks for sharing this wonderful tutorial.

[@AnaCristina-qe8hx]

Muy buen trabajo, mas videos en español por favor. GRacias :)

[@aparnadinesh226]

Hi I have a doubt in credit card fraud detection model 1 Precision, which is actually 0/0. how is it 100%. can you please clarify this