entropyAndGain

[@erythsea]

This is the BEST video on tree, gain and entropy on KZbin.

[@Madeinchinaagain]

Very helpful. Videos like these make procrastination and cramming possible.

[@ArminBishop]

I own you my exam. You are my saviour. Best explanation ever!!!

[@keithmaly904]

Very nice walkthrough of the concepts of entropy and gain. Ty.

[@davidinfante4887]

Good explanation. I am going to give a diplomat on data science to several colleagues (I am the head of the group of embedded systems with expertise in data mining, visual anlytics) and I was looking easy examples to explain them. This is the example with 0 entropy and highest gain I have found. All the best

[@JoseFerreira-dk3gv]

Good video.. in minute 22:50 you pick temperature as the one with maximum information gain (0.570).. shouldn´t you pick Humidity (0.970)? I know that in terms of entropy you want to pick the one closer to 0.5... but that one refers to gain..

[@aujoeyd]

U probly right

[@snehanshus7039]

you're right..max gain should be picked....sleep of tongue perhaps

[@sivakumarrallabhandi4548]

That's correct. Excuse the typo and appreciate the video.

[@channagirijagadish1201]

Clear and Concise Explanation. Appreciate the good work.

[@fiacobelli]

Entropy is caos. Not knowing what decision to make. An entropy of 1 means that with the information you have, you can't make a good decision. For example guessing what side is a con going to fall has full entropy. You don't know. Now, if I tell you that the coin is loaded and it falls on heads 70% of the time, there's less entropy because the information you have allowes you to make a more precise guess.

[@KumarKumar-fc2gc]

Thank you so much for the crisp and clear explanation....

[@harisk.1466]

21:29 I think you should say, you choosed "outlook" because it has highest information gain and the lowest entropy among all other features? Correct me if I am wrong!

[@cdm891]

21:30 where does the entropy of .970 come from? Gain(sunny, Humdity) = .970 - (3/5) …. etc where did you get the .970? Thanks

[@arjunsrinivasan3751]

It's the entropy of the set Sunny. All Gain examples are for sunny

[@datascientist2958]

Can you guide Gini index work with information gain to rank feature? Or Gini index work with Gini impurity to rank feature?

[@kulin6522]

Really good video, congratulations to the creator

[@vyshakhunnikrishnan]

22:40, you calculated all the gains correctly and didn't choose the highest gain attribute. Constructing a tree in this fashion would mean it is deeper and more complex when we are looking for simplicity.

[@fiacobelli]

Vyshakh Unnikrishnan correct. I should have picked humidity. My bad.

[@fiacobelli]

Correct. I picked the wrong one. I should have picked humidity

[@IgraphyRupage]

This video is great! Thank you!

[@pruthvirajsuresh3467]

crystal clear. thank you very much.

[@OskarCeso]

Made it clear for me. Thanks!

[@cdm891]

im really confused if someone can please help me understand.... first you get the entropy each attribute, then the entropy of each subset of that attribute, then you find the gain....but to construct the tree you look at the entropy of the attribute not the gain? what is the purpose of finding the gain if you dont use it to construct the tree? Thank you

[@snehanshus7039]

you use the gain to split the tree

[@pericharlasai7242]

What is entropy

[@LutfarRahmanMilu]

This is a great tutorial! Thank you!

[@johnstyl]

21:06 you forgot to say to also subtract the P of negative result multiplied by the log base two of that for the entropy of high wind, great vid aside from that

[@cyrilchubenko2657]

Thank you very much! At last I get the idea of those variables

[@sidk5919]

Thanks, you made it really clear!

[@melkyewereta8113]

I found it very useful..thank u.

[@sukursukur3617]

Wauwwww