Xgboost Regression In-Depth Intuition Explained- Machine Learning Algorithms 🔥🔥🔥🔥

[@krishnaik06]

121/2 is 60.5 :)

[@lokesh542]

144/5 is 28.8

[@nihalshukla7718]

243.42

[@saitarun6562]

haha yeah just now noticed and came into comments

[@saitarun6562]

@@lokesh542 haha yeah

[@saitarun6562]

@@nihalshukla7718 yes noticed

[@vishaldas6346]

Sir, I'm a huge fan of yours, although I know Xgboost for regression and after watching this, I can say how simple this is. You clearly explained each and every concept like Similarity weight, how to make a split & Gamma for pruning. Unlike other youtubers who've made this algorith complex, now I can suggest my collegues this video.

[@shubhammore5084]

Please make practical implementation....much needed and its gonna be amazing!

[@mdmynuddin1888]

avg(-11,-9) will be 10?

[@alishazel]

I like this video as among alll videos i can understand your accent i hope you can redo the video .....

[@raneshmitra8156]

Eagerly waiting for the video...

[@mat4x]

For the similarity weight of the root, the square will add up to 405. You just cancelled them all?

[@ronylpatil]

Very clear and understandable explanation. Keep posting and keep growing.

[@gauthammn]

Very nicely explained. Thank you. Had a quick question. Why do we not use the similarity weight to determine output in regressor xgboost? In classifier xgboost the output is based on sigmoid of similarity weight

[@phanik377]

One question: Is it sum of Residual and Square. (or) Sum of Square of residuals ? I think it should be sum of Square of Residual. Which mean we need to Square first and then sum

[@juliastelman4189]

I also had the same question

[@ashwinshetgaonkar6329]

nice implementation explaination, statquest + this tutorial is a very effective combination to grasp this concept

[@gouravnaik3273]

so xgb and gb are similar just the approach for tree creation is differetn in gb we use entropy or gini for information gain and in xgb we use similarity weight for information gain. with some added pruning facility

[@rohanthekanath5901]

Hi Krish, Could you please make a similar video with regards to working of catboost and lightgbm

[@krishnaik06]

Sure

[@rohanthekanath5901]

Those videos would be great as there is nothing available like that on youtube

[@NaimishBaranwal]

In the output value's formula regularization parameter should be added in the denominator.

[@tejas5872]

Please create a playlist on reinforcement learning

[@bill-billy-bo-bob-billy-jo2573]

Krish, RockStar of actually teaching

[@shashvindu]

I am waiting

[@rafsunahmad4855]

is knowing the math behind algorithm must or just knowing that how algorithms works is enough? please please please give a reply.

[@rishabhjain1418]

This video is strikingly similar to StatQuest's ....

[@suganyasuchithrra6992]

good morning sir...can you please share LGBM algorithm....

[@lol-ki5pd]

so only one column per decision tree?

[@Trendz-w5d]

Idk why I'm not understanding this splitting. Why you create output of all on behalf of just one split?

[@sravanakumari3626]

sir while creating the tree every time from which feature we have to start from. is there any metric for that .

[@pawangupta8948]

How did you know which root feature to take?

[@samratsakha4274]

Then. whats the difference between XGBOOST and GradientBOOST sir ?

[@Niteesh-v6c]

I don't see an intuition here. This explains how it works.. I was mostly looking into why it works

[@nizarscoot2844]

please do self organizing map i had an exam after 3 days and i failed to understand it

[@ahimaja2261]

Thanks

[@puleengupta3656]

When you changed 41 to 42,Average also change

[@nakuldafale1477]

The average is 60.5 not 65.5

[@shantanusingh2388]

121/2 is 60.5 i know its not a big mistake but sometimes i take notes from your video and while revising after a month if the values are wrong i need to do the calculation again and it also creates doubt

[@khubeb1]

How you are selecting < 2 and > 2 ? Please clarify

[@MuriloCamargosf]

In the similarity weight computation, you're squaring the residual sum instead of summing the residual squares. Is that correct?

[@krishnaik06]

first we need to sum and then square :)

[@burakdindaroglu8948]

@@krishnaik06 Are you sure? This contradicts the formula you have for the similarity weight.

[@avinashajmera80]

similarity weight you have written fromula sigma (x square) while you are doing square ( sigma x)

[@santoshhonnungar5543]

Lots of mistake in this video Krish

[@arjundev4908]

You can ignore aggregation mistakes. But steps are correct.

[@WhittierElliot-f3b]

White Jennifer Rodriguez David Jackson Angela

[@rwagataraka]

Thx. Waiting for the video

[@WhittierElliot-f3b]

Jackson Michael Lewis Larry White Matthew

[@samsimmons8370]

Are the similarity weights (sum(residuals))^2, or sum(residuals^2)? Those end up being very different numbers. You initially wrote sum(residuals^2), but implemented (sum(residuals))^2

[@abhisek-chatterjee]

Krish,can you tell me about some references for gaining in depth theoretical knowledge about various machine learning and deep learning models?I am currently pursuing masters in Statistics,so a good chunk of them comes under my syllabus.But things like NLP,DL or xGboost,recommender systems etc is not included.Anyway,your videos are great to watch.

[@anuragshrivastava7855]

at 12:36 u have calculated gain which shd be 243.58 bt u calculated 143.48

[@pranavreddy9218]

how can we consider first prediction as average, in xgboost regressor using Scikit learn, we see 0.5 as initial prediction, how to change this 0.5 to average value, can u please ML model with same data

[@AnandPrakashIITISMDHANBAD]

Thank you so much for this wonderful session, one silly mistake in the video is 121/2 = 65.5, remaining contents are okay.

[@abhishek_maity]

Finally this one :)

[@sidindian1982]

19:20 . How The Gamma - Value 150 .. is set ??? Who assign this ???

[@pravinshende.DataScientist]

I feel xgboost too much complecated so i chose this vedio of krish naik sir because he make the things very simple so .. and now lets go .. Thank you sir very much !!!

[@deepkumarprasad6277]

at 14:17 you output will be average but you say again 20

[@priyadarshinigangone2490]

Hey can you please do a video on XGBOOST Regression implementation using Pyspark

[@atomicbreath4360]

Sir what exactly is difference between base model trees created in gradient boosting and xgboost.?

[@gauravverma365]

Can we generate the mathematical equations between adopted inputs and output after successful implementation of xgboost?

[@RishikeshGangaDarshan]

You are so good

[@vatsalkachhiya5796]

Hi Kris there is an issue with the similarity formula it should be " (sum for residuals) squared/ number of residuals+ lambda" you have written "sum( ( residuals) squared)/ number of residuals+ lambda".

[@mohammadyawar2016]

Hello Krish! Thank you for making XGboost extremely easy for us :P I have a question: Is that alpha or lambda that you refer to in the similarity weight equation during the lecture?

[@kdmyt8709]

Please make one video on in depth intuition on Gradient Boost Classifier Problem.

[@anmol_seth_xx]

After watching the XGBOOST classifier video, this lecture is a bit easy for me to understand. Lastly 1 query, I have, i.e. Till when we have to repeat this XGBOOST REGRESSOR process?

[@shivanshjayara6372]

@14:06 how output could be the average.....average is taken only for the base model only?

[@vishalpateshwari]

Can I get more info on Feature Importance calculation and regularization?

[@karthikeyapervela3230]

@Krish thanks for the video! If two features have the highest gain and both gains are similar on what basis Xgboost choose which feature to make the split on?

[@RishikeshGangaDarshan]

Sir please make a video where we start the process for solving a problem and where we use this technique like pca visualize xg boost Fearure selection , logistic regressions svm liner regression etc Then we can easily understand the the right path for solving the problem because we have read more things but confused where I start and what use for problem solving please make a video sir

[@sandipansarkar9211]

Greatvideo.Veryvery important to gain success in product based companies

[@sparshgupta2931]

Sir, is this video enough for Interviews? Like if I have applied XgBoost Regressor to a project & the interviewer asks me to explain the algo.

[@nishiraju6359]

Nicely explained .. keep uploading more n more videos .. @Krish Naik Sir

[@v1hana350]

How can parallelization work in the Xgboost algorithm? Please explain it with an example

[@chrisogonas]

Well illustrated, Naik! Thanks 👏👏👏

[@MittalRajat]

Kindly send a new discord link. It has expired.

[@lakshmipriyaanumala7331]

Hi sir, Can you please make a video or provide with some insights on how to get research papers on deep learning

[@inderaihsan2575]

thank you very much!

[@MittalRajat]

Your discord link has expired

[@iftiyarkhan7310]

please deploy one model in fast API

[@marijatosic217]

First of all, thank you so much for everything you do here on KZbin. I do have a question, why is the base value sometimes the AVG and sometimes we calculate it by using the Loss function and finding the first derivative?? Thank you! :)

[@marijatosic217]

We actually get the same result, so never mind :D

[@manojrangera]

In both case we will get tha same result and that will be average of output in regression so.. We can use average in every situation (regression)

[@jamalnuman]

really great. one of the best explination i've ever seen

[@madhusriram2860]

Excellent

[@ajaykushwaha-je6mw]

Best of the Best

[@devaganeshnair5883]

Thanks sir

[@morrigancola6154]

Hello! The Res 2 will be computed by the difference between the Res1 and the predictions made by the tree 1, right?

[@durjoybhattacharya250]

No. Base Model minus O/P. Target is to minimise the Res n as n increases.. with constraint till model doesn't overfit.

[@chenqu773]

The moment when your wrote 20/2=10 (instead of -10) as the gain of left branch, I realized what means "gradient exploding" :D:D:D Many thanks for these awsome tutorials !

[@divitpatidar8253]

can u please explain i didn't get this part brother

[@vatsalshingala3225]

can you explain it bro