Performance Metrics On MultiClass Classification Problems
HTML-код
- Опубликовано: 8 фев 2025
- 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
Please do subscribe my other channel too
/ @krishnaikhindi
Connect with me here:
Twitter: / krishnaik06
Facebook: / krishnaik06
instagram: / krishnaik06
![Felix "Unfair" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/Oswujxm2Ag0/mqdefault.jpg)
Guys, lets applaud his efforts. This video is being shot at 12 in the morning :)
I follow ur machine learning videos & it helps me a lot. Could you please make a video on multilevel multiclass performance metrics (absolute-true,absolute-false,accuracy,coverage,aiming)? I can't find out any appropriate source for it! Please try to do that!
To the point video. Awesome Krish. Your videos save hours of efforts for many students.
That was a superb video.Got all my doubts clear.If you could the ipynb notebook or google colab for practise would be great .Thanks
in the report
suppose:
pre recall f1-score support
a
b
c 0.0 0.0 0.0 3
what do mean by the above report
it has an accuracy of more than 0.6
remaining two classes predicted the precision and recall score with support
Sir, how do I know which row and which column refers to dog, cat or fox? I mean how to know row 1, 2 or 3 refers to a dog?
What is correct confusion matrix , now I am confused , In some places FP become FN & vice versa
Thank you Bro
sir you have explained Accuracy and precision pretty well but Recall and other performance matrix are not explained only cursory introduction is given.
Hey , for recall you just have to see the correctly predicted cats in y_pred and total number of actual cats in y_true. So according to data correclty predicted cats = 4 and total actual cats = 6 , so 4/6 = 0.667. Similarly for dog and fox.
Ya that's it. I hope it was useful.
Thank you... :)
For multiclass classification which parameter will be more important precision or recall and why??
Thanks Krish
Thank you again !
simple question:
You said when we have imbalanced data set , we cannot always go for accuracy, we have to consider precision or recall to judge the accuracy. Now the above final matrix which we have how can we find accuracy here, is it the F1 score we should consider, but there are 3 classes. I am a little confused here
i have not received any answer of my previous question, looks like krish needs to hire assistant who cam answer, we all know how busy he is.
I will have another question.
if i have 3 classes : 0,12
i create Confusionmatrix for all 3 lest say accuracy for 0 is 98% for 1 is 66% and for 2 is 65%. considering my Training data set is equally distributed in classes, can we say model is biased for class 0
@@aashishraina2831 most questions in comments go unanswered, hope it helps you - towardsdatascience.com/confusion-matrix-for-your-multi-class-machine-learning-model-ff9aa3bf7826
@Krish Naik. If we have too many classes, then I think confusion matrix will create confusion and not be that much easy to interpret. Sir any other performance matrix for multiclass classification except confusion matrix?????
Great Sir.
Hi what is support in the classification metrics report and what its equation also,what are micro average and other values in the below of report.
@krish naik,instead of using confusion matrix on multi class , can we use Kappa score as performance matrics on multi class
Yes we can..Kappa score should be looked at when uu imbalanced dataset
@@krishnaik06 hi sir, on which dataset (test or is it validation?) We should perform this mterics
Is it possible for precision, recall and f1-score to be zero? if yes then what does that mean
How do I plot the confusion matrix for multiclass image classification(say 4 class) while using ImageDataGenerator for augmentation in CNN algorithm?
sir could u plz explain multi class classification with random forest
How to calculate sensitivity and precision for multiclaa
Thanks Krish for the explanation. But the github link is missing .
Sir, please make a video on ROC and AUC with practical example..!!
Sir can you please upload video on indepth intuition of logistic regression cost function. It’ll really help to understand concept in better way
Krish, please upload further Deep Learning videos.
Hello Sir,
I am doing coding for the breast cancer dataset with Keras Sequential Model.
while printing classification_report and confusion_matrix it says the following error
''Classification metrics can't handle a mix of binary and continuous targets'''. Previously, I was able to do the same, but now showing this error. Kindly help me out.
Sir plz start classes on 10th standard Artificial intelligence plz
can you do an explanation of the DAG mumlticlasse scheme used on SVM
Sir please explain neural network technique on multi-class problem with imbalanced dataset
Sir can you explain Multi class student's performance prediction based upon on machine learning
Dear sir, I am using Rbf. I want to know how to select the hyper parameters. please kindly help
Hi you calculated sensitivity ( #_of_ TP/#_of_TP+#_of_FP). But why scikit learn mentioning as Precision.
What is support?
Do you kown wgat is support???
What is mean avg precision?
how to find confusion matrix for model used in semantic segmentation??
can you explain how to plot ROC curve using Image Generator(flow from directory) in classification multi classes?. Please
How to calculate "Specificity" here? Can someone please help?
Hi Krish , Please upload the GitHub link for the above video. Thanks
TypeError: Singleton array 2836 cannot be considered a valid collection.
please upload the class file.
It looks like FP and FN are interchanged. Sir, I'm confused as your previous video on binary classifier performance metrics - ruclips.net/video/aWAnNHXIKww/видео.html shows differently. Also didn't understand the precision calculation for cat. it should be row wise not column wise addition on denominator right ? My understanding on precision is Out of total positively predicted cats, how many are actually cat
sorry , i'm wrong. You have interchanged the actual and predicted class in the confusion matrix.
@@charlsjoseph It happens, i had the same doubt!
First like
..