Tutorial 41-Performance Metrics(ROC,AUC Curve) For Classification Problem In Machine Learning Part 2

Please join as a member in my channel to get additional benefits like materials in Data Science, live streaming for Members and many more
/ @krishnaik06
Complete ML Playlist: • Complete Machine Learn...
Please do subscribe my other channel too
/ @krishnaikhindi
Connect with me here:
Twitter: / krishnaik06
Facebook: / krishnaik06
instagram: / krishnaik06

[@skviknesh]

The explanation is great! Why is it used? 1. ROC & AUC - Plotted between TPR & FPR, helps in - visualization & explanation & selection of a required threshold for the model! What is ROC & AUC? 2. ROC - (Receiver Operating Characteristic curve) that you have drawn & AUC is the - (Area under the curve). How does the curve look like? 3. AUC should be greater than the 0.5 line drawn, which indicates a better model. 🙇‍♂️🙇‍♂️🙇‍♂️

[@suhaillone831]

This guy is definitely one of the best teacher available on KZbin Simple but effective explaination Lots of love for you Sir ♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️♥️

[@LIMLIMLIM111]

This channel is a gold mine. Thank you for your knowledge Krish.

[@arjyabasu1311]

Indeed

[@priyankrajsharma]

Agree

[@borhanmukto7770]

Bro, You are a gem. I mean not a single unnecessary word, everything is explained clearly and concisely. Many many thanks brother

[@guillermotorres4988]

After reading about roc and auc, your example calculating manually the values was perfect to finally understand this topic. Thank you!

[@ayberkctis]

Hey Krish, you are amazing! You explain topics clearly. So, everybody can understand it!

[@hassamdaudi]

I swear to God man, I learn more from you than my professors at school... you're saving my assignments

[@DappuDon]

In which school we are teaching ML?

[@mandeepsinghnegi1931]

Sometimes, I check this, that and then just search if you have made a video on the topic. No one can just simply explain better. You are an AVENGER.

[@230489shraddha]

Krish you explain data science concepts so clearly that I have become fan of your's !!! thanks a ton man!!!

[@sandipansarkar9211]

Thanks krish.My doubts about ROC ad AUC are now clear.

[@yashodhansatellite1]

Hats off Krish..u r so deep in knowledge

[@smiling_madly596]

you're a true supervisor for supervised learning! 🌷

[@chaytanyakumar8939]

Best explanation on the internet. Thanks Krish!

[@teamannonymous163]

totally clear .. you way of explanations are really very amazing.

[@mohammadarif8057]

Sir people like you should be admired ...big fan hope we meet one day

[@tengliyuan1988]

Krish I really appreciate your effort in sharing your knowledge. Thanks

[@KiranKumar-bb8lr]

really ur communication ur voice and the way u explain it stole my heart broooookeep rocking broooo

[@athanikarammy7476]

great explanation video..anyone confused or overwhelmed about how much to study and from which channel to study...simply follow krish naik playlist from start to end.....

[@narenderbansal5741]

Very nicely explained. Thanks for sharing.

[@sumitnaina]

You explained it so nicely and make me understand the roc and auc concept so easily. Thanks a lot krish

[@syedyunus4204]

Krish, please make a video on "implementing all the Metrics For Classification Problem in ML by taking improper data set" as you mentioned in one of the videos of ML playlist (#1:05 minute of Tutorial 34 of Complete ML playlist)

[@happytorev466]

Thank you for this video. Eagerly waiting for part 3.

[@xyzaex]

Brilliant. You are officially Sensei.

[@hilaav7449]

Hi I couldn't find anywhere part 3 - performance metrics for classification problem part 3 ( you said in the first one there are 3 parts I only found 2).By the eay, I became a memeber more than half a year ago to support your work, because you are an excelent lecturer and you helped me a lot.

[@avinashmandora8745]

What an informative video. Really amazing!

[@agenthitman7564]

Finally I have understand ROC....Thank YOu Sir

[@hemrajbhatt1034]

Thanks a lot sir. I was seriously waiting since your last video on performance metrics.

[@MrYoonyan]

The best explanation ever seen, thank you

[@binnypatel7061]

Keep up the good work.....Great explanation!

[@mr.top5636]

Your explanation is just amazing go ahead man

[@chandrusaekarb3445]

Nice Video and clear explanation. Thanks a ton ! .. Please do post part-3 of the metrics soon....

[@chubiandchuba4804]

Hi Krish you are really superb... Excellent knowledge, well explained

[@saikatkar547]

Very nicely presented ...clearly understood . Thank you sir

[@lavanyathippineni6341]

Superb explanation👌👏👏👏....Waiting for part-3. Please upload it as possible as sir

[@agnelraju718]

superb explanation loved each bit😍😍

[@ianmutuma6564]

Thanks Krish. This saved me a couple of hours

[@souvikmaitra6653]

awsum explaination...ALL THE BEST for all your upcoming videos #Krish.

[@rishigupta2342]

Brilliant explanation. Please make part 3 of this video.

[@firasfakih3050]

I just hit the jackpot with this channel, thanks alot

[@sharinaathikkal]

Me too

[@lumar_music]

Please consider using manual focus and iso :D

[@kranthikumar7635]

really awesome explanation....

[@fortask5897]

Lovely Explanation.

[@sibamarcel9428]

You have great knowleg...Thanks a lot

[@amoldeshmukh8244]

Hey Krish , Did you uploaded Part-3 of this video (i.e. Performance Metrics for Classification problem) ?

[@adityagandhamal2351]

Thank you sir....It was very helpful.

[@kaustubhshete6250]

Awesome Work Sir !!

[@dordonduty2918]

very short and srisp, i love it

[@ozcanf6680]

Great video, thank you Krish

[@abhishekranjan2617]

Awesome video sir. I love it😍😍🥰🥰😘😊😊🙏🙏🙏🙏🙏👌👌👌👌

[@sohinimitra7559]

Hi Krish, your videos are very informative and helpful. I saw your video where you explained ROC-AUC based on thresholds in Logistic Regression. What is the intuition about thresholds for other classification models? For example, a decision tree will split based on feature value to determine the class, how is ROC constructed here? Similarly, for other classification algorithms, how is ROC constructed?

[@fancy4926]

Krish, at the end of this video, you said: "if they focus on both TPR and FPR, we can also choose TH(0.4) at that plot". But in real ROC there are always a lot of inflection points like TH(0.4), then which inflection point should we choose?

[@jagadeeshabburi570]

Maza agaya bhai dekeke. Subscribed. The thing what I felt lacking was. More explainanation

[@sampsonagudze8674]

Fantastic!

[@pranjan85]

Hi Krish, Can you please upload Part-3 so that everything can be understood

[@myGiG-09]

Awesome lecture

[@abhishek-shrm]

Great video Sir.

[@sonalighadage6253]

very nice explanation

[@ankita684]

Hi...can you share the link for part 3 of the performance metric video. Also, can you share any real life example when domain expertise would want threshold to be say 0, 0.6.?

[@tuannguyen4802]

Amazing explanation

[@rajeshjha9246]

sir, please give one walk though session on classification problem with all Performance Metrics(ROC,AUC Curve,precsion,recall etc.)

[@narenderbansal5741]

Hi Krish, Suppose i want to set the threshold at .75. then how i can do this in python code. is there any parameter in ruc_curve method or have to do it by using logic/code.

[@preetigupta1349]

I think Krish, if you can include Gini index computation part that would add lot of value to this part of series.

[@hemantdas9546]

Super Sir. Very clear

[@anujlahoty8022]

Awesome video.

[@pshubhamoy21]

nicely explained

[@louerleseigneur4532]

Thanks Krish

[@adiflorense1477]

8:20 here I finally understand how to read the ROC curve. You really are great

[@sanketannadate4407]

it really helped thnx

[@krishnabhadke6161]

perfectly explained

[@sahithim9278]

for the threshold value of y^=0.4 and y^=0.6 , tpr and fpr values is same i.e.,(0.5,0.5) and for y^=0.8 is (0,0.5). Now the graph will be different and AUC might be less than 0.5.

[@hussainlokhandwala2245]

Yes!

[@santanusarangi]

How to reset the threshold value?

[@subhasrik1436]

sir excellent ,plz explain eer from det curve

[@mohsin10185]

Excellent

[@shaz-z506]

Good video Krish, could you please create a video on Cohen-Kappa score.

[@megalaramu]

Hi Krish, Awesome explanation. Thanks a ton. Could you please help us by giving an example (Part-3) for all the performance metric in python.

[@kanakkumar5261]

Hi Krish, awesome teaching skills... please upload all parameters in python programming language take an example like iris data

[@avisii8928]

hello sir, i am working on my final year project, Resume classification and ranking system using knn and cosine similarity, the roc auc curve for all my class label came 0.5, what should i do?

[@kaisersayed9974]

Sir please make videos on web scrapping for beginners.. please sir and I am anabel to get the link of AQI website which you have shown in the video ....but please make more videos on web scrapping..

[@tannurohela6192]

Great explanation but I didn't get how to take the threshold values , this is , the values we have taken to predict and draw ROC. Can we select them randomly or is there a way?

[@asawanted]

So comparing the Andrew Ng video, the true positive rate is same as recall? and the false positive rate is same as precision?

[@godse54]

krish please make part 3 and also list some performance matrix for multiclass classification.

[@mounicak5923]

Great videos

[@sushilchauhan2586]

CAN YOU PLS EXPLAIN MULTICOLINEARITY AND HOW TO AVOID IT ? PLS or PERTUBATION TEST

[@tomasvonbischoffshau]

Thanks! really.

[@chitrashivraj8403]

You are awesome krish. It takes so much effort to reach here. Brilliant :D

[@adiflorense1477]

0:52 how to determine the threshold. can we set the threshold ourselves?

[@akashs9414]

Actually, many will sort those probabilities in descending order and use them as thresholds.

[@ravinch7973]

Explain about epoch and its graphs with accuracy, Precision, Recall, f1Score. Also show in python or matlib how to draw

[@RomeshBorawake]

Hi Krish, Thank for the video. For calculation AUC score we have to sort y^prob or threshold list in accending order?

[@kautilyamisra9231]

true

[@neelroy4153]

Hi Krish, how did you come up with original y hat?

[@iamramanr4s]

In the case of binary classifier that outputs 0 and 1 labels how can we calculate auc score and roc curve

[@sanskrititiwari4267]

Your channel is very informative. Please bring part 3 as soon as possible :)

[@alakshendrasingh3425]

please show the implementation of roc and auc in python

[@avinashreddy5779]

thanks for your knowledge Krish sir, I am new to date science ROC used to evaluate the logistic regression only or do we have any uses can anyone help with this

[@amitbudhiraja7498]

Can someone tell that as the sir said we have to see all these different metrics when we have an unbalanced dataset if we have balanced datasets then only accuracy is enough?

[@dheerajkumark2268]

Sir can you plz make videos on statsmodal and scipy library

[@sonamkori8169]

Thank you Sir

[@ektamanvar4397]

waiting for part3 video for performance metrics for classification

[@salsabilemed6700]

thx a lot

[@saurabhbarasiya4721]

Great

[@novinnouri764]

thanks

[@saivenkat8813]

hi krish in the previous lecture u said fpr =fp/p =fp/tp+fn now in this video it is fp/fp+tn which one is correct