Machine Learning Algorithm- Which one to choose for your Problem?

[@nandinidasgupta7781]

This is the only channel i guess who understands what students are expecting to know other than traditional algorithm knowledge.. Thank You so much..

[@hunainaghai3342]

Which algorithm did u choose, Nandini?

[@srikanththecoolhunk]

You have no idea how useful all your videos were to help me find a job. I cannot thank you enough Sir, please post more.

[@varunsagartheegala]

Your videos is a gift to all data science aspirants like me and working professionals too. Thank you for making our learning easy and fun. Please don't stop

[@harshj84]

Hi Krish, I really enjoyed your videos. The specialty of your videos are they fill the gap between theory and practical. I have watched tones of other videos, most of them are telling "how to plot the graph" but no one is explaining "how to leverage information from these plots?" Just like this video, we need to use knn as most of the points are overlapping. I would really love to such a great video for other algorithms as well. Not only me, but I believe your subscribers would also love them. Tons of thanks to you.

[@anandacharya9919]

This is your best and most important video of all. Thank you 🙏

[@chravi511]

Hi Krish, I've been following your channel for most of my learning. The way you explain the things gives a very familiar approach and giving out what is needed in the way everyone can understand. This let me watch any video of yours till the end and learn completely. Happy to learn from you. Thanks

[@deepuraju2456]

Hi sir , can u give any idea problem statement for loan approval prediction

[@lisamathur9206]

Ssly this is exactly what I've been wondering from so many days. Weeks actually. But didn't find even 1 proper Video on this . Thanks a lot......

[@nokiyem521]

Great, this is the video i was looking for to explain the difference from a basic mathematical perspective.

[@akashchoudhuri3932]

Best video till now. Thank you Krish.

[@puneettiwari2251]

Great session, Please make an elaborative video on this topic covering all the pros and cons of different algorithms and if possible with codes. Thanks for this session again sir🙏👍👍

[@rohinikalamkar5736]

This is what is required .. thank you so much for sharing this 👍👍 great work

[@ankitaB11]

Most important video. Explanation is just amazing.

[@pravinkumar54]

@Krish Naik 11:30 if you double pick the pairplot in jupyter notebook. It zooms in

[@gargidixit5651]

Amazingly Explained, I have never got any satisfactory answer for this question Thank you so much for such knowledge sharing :)

[@sahinur_zmn]

just awesome 💚💚💚i am just recharged after completing this video..thanks a lot.

[@izzycrux48]

Thank you Krish for the amount of effort and insights you put into your videos. Really helps a lot🙏❤️👍 May God bless you and keep you well🙏

[@vikasbohra3253]

You are AWESOME Krish not only from knowledge's point of view , but also in explaining the concepts in pretty well manner. Thanks.

[@sambhavmishra1873]

Understood it very very clearly ❤ whatever doubts that were arising was getting explained in the next second.

[@chravi511]

Please Explain me, Scenario 1 : All Categorical, ordinal, nominal features - Categorical target Scenario 2 : All continuous features - Categorical target Scenario 3 : Combination of categorical and Continuous features - Categorical target Which model to use in these scenarios (particularly SCENARIO 1) ?

[@amoghjain]

Such a practical channel with real world applications! Thankssss!!

[@neelammishra5622]

I really found the answer of my most confusing question into so simplify terms. Thanks sir

[@nagupranav]

you are always great man.. anyone can understand by watching your video.

[@translatethis7765]

The thing is, you are looking at the pair-plot for only two dimensions. Although everything is overlapped when projected into pairwise space, it doesn't necessarily mean the data is not linear separatable right?

[@himanshubisht9643]

Here I can understand how to choose the best algorithms to my dataset thanx sir👍👍

[@saunitmarolia3901]

Thank you Krish! All doubts cleared now.

[@aanalpatel5361]

Most of my questions were answered through this video!! Thank you a ton, sir!!

[@thongtech1984]

thanks a lot for this wonderful explanation, Krishna. You are my new hero now :)

[@MrDeepinaction]

Great video, friend!! Keep up the good work. Need to learn a lot through you.

[@khalfanalkharusi6577]

Thank you for your contribution. Please note also that classification learner app in Matlab provides you such pair plots

[@itybahadur3625]

Excellent video for an initial understanding! Thanks!

[@satriogani3105]

In the pairwise plot, the points are only plotted on two dimensions. However, our data have dimensions more than two. This means that we do not plot the other features. It looks that we do some projection from higher dimensional space to two dimensional space. Can we conclude if the groups overlap with each other only by looking on two dimensional plots? Simple example, if we project the earth on two dimensional space, we can see that the norther and southern hemisphere will overlap on each other.

[@translatethis7765]

I have been looking for comments like this... I agree with you. I don't think one can conclude whether the data is linear separable, from just looking at pairwise plots.

[@navya-s3v]

Selecting the appropriate machine learning algorithm is determined by aspects such as problem type, data qualities, and desired results.

[@abhinav02111987]

You saved my day Krish. Thank you.

[@sankarapandian.selvaraj]

All of your videos have a lot of useful information. Thank you...

[@roshanajadhav5854]

You are doing amazing job Sir, I am too late to land on your channel Keep it up You are providing knowledge free of cost where people taking money for this in thousands No need to join any private classes with huge pay if you will provide this material in sequential order Great job all the best👍💯

[@fuck30093]

Have you found an answer to your question?

[@chrissanjiv1753]

good explanation Krish, Very crisp and clear

[@anandhasrivi]

Best way is to try logistic regression right away and result is poor. Then we can conclude data is not linearly seperable. Knn will not be suitable for larger dimension data due to curse of dimensionality. I expected more from this vedio but vedio is very simple

[@yogitapeshane6934]

It was really good video.. Please make more videos in DataScience...

[@timetimetime3985]

If you want to predict the future or get the information for future use regression if you want to distinguish the data from other data use KNN, Random Forest etc, that collect BIAS

[@ranjaniselvaraj6475]

Good Job! Clearly explained. Thank you so much for this video.

[@swetapatra]

ok, so ideally, we decide on the algorithm based on the charts we have plotted?

[@V_Wankhede]

Informative video , its all jam of each algorithm that you taught. Thank you for brief explanation. 👍

[@devmaharaj4640]

One of best video's I have seen , won't forget to return u the favour....

[@learbasicsofpythonwithus-pe1kk]

I never saw a video like this description...

[@kamran_desu]

Hey Krish, great video - just shows how important it is to understand the geometry/maths behind ML. In regards to using KNN here for overlapping data points, would SVM (radial/polynomial) be a good choice as well, since it will use the kernel trick to apply a non-linear classifier in a higher dimension? And how would SVM compare to the tree methods in terms of computation complexity?

[@adityakishan589]

Same doubt. Krish please help us get clarity on this.

[@karandua6564]

Even logistic regression may work, he is just comparing two dimensions at a time, they may be linearly separable in higher dimensions .

[@danielschulz5447]

@@karandua6564 no, I think that is not true. LR will not work good, if single plots are pairwise highly overlapped and/or not dividible into straight lines.

[@harshinimedha7807]

The best explanation I have ever seen ✨

[@souravbiswas6892]

awesome work Krish, a big thumbs up

[@olajideoluwasanmi4260]

You are the best Krish. Thanks for this.

[@mansiparikh8357]

You have explained in a great manner. Please make more such videos for data science. Very helpful

[@chandramoulireddy9636]

Naik sir, thanks for giveing inforamtion about selecting algor,,. i am very happy to follow ur channels. thanks.

[@somalkant6452]

hi krish, it was an awesome video. just a doubt popping up in my mind, when there is overlap of datapoints, why cant we use SVM, bacause that will also take the points (of different classes) to other dimensions and divide the points using hyperplane. please correct me if my understanding is not correct. Than you so much for your awesome videos.

[@chinmayjape4981]

Please make a video on how a beginner can participate in Kaggle competitions.

[@ritwickjoshi]

brother you need to dive in to find out Dont worry about the result see where you lack and work from it but remember not to go down from your baseline

[@siddhantkaushal7669]

It is a good video. Also explained in detail to understand pair plots. :) Thanks.. My Question is: When should we decide to use SVM and Naive by visualizing and analyzing the pair plots? :)

[@datahat642]

@siddhant Naive Bayes works well with higher dimensionality (such as text corpuses) while support vectors whpith its kernel is applied when the data is non linear but you want to apply a linear classifier by transformation

[@faisalghazanfar169]

excellent explanation...really an easiest way to understand and clear confusion..:)

[@amirhosseinrahimi3964]

Thanks a lot for such a great explanation 😊

[@vijaynale7893]

Nice ... Please make more videos for data science, beacuse it better to understand into that way...... Thank you Also make video for data cleaning, Solve any business problem

[@pietart3596]

Thanks for everything you do, your words are very motivating

[@gouravgangani]

Thanks Krish for such a nice explanation.

[@theindianrover2007]

Liked this video, but in case of large test set KNN might prove you costly at the time of prediction

[@MyTube6686]

how ? can you explain further Please

[@Rajkumar-vc2pg]

You are my true datscience Guru 🙏🙏

[@leoprinceznirp39]

Thank you so much. Your explanation gives good clarity. Great work. Thank you😃

[@KaranSingh5]

Thank you so much for this video! This solved such big confusion for me!

[@snehamonga2004]

You are the best teacher

[@karandua6564]

At 13:30, you are making an assumption the data is not linearly separated based on just two dimensions. That's not correct, since the data can be linearly separable in multiple dimensions.

[@saichoz]

Great Vid. You made things clear. Thanks

[@cseswadhinpanda]

Nice explanation , you explained us like teaching a parrot love you from india .

[@ninjaduck3534]

This was so excellently explained, thank you so much!!

[@KetanChaudharyTHE-GREAT-KETAN]

Thanks a lot Krish......it's really very helpful ......

[@asankab.l.5166]

Your book is out of stock, please provide a way to buy. Great tutorial!

[@rajivturkar1671]

Great explaination Krishna.. I would like to know how pairplot will show categorical variable and how we will do the feature engineering for categorical variable?

[@sejalchandra2114]

Yes, this is my doubt also. Please help!!

[@manikaransingh3234]

@@sejalchandra2114 You first have to do the label encoding to your categorical features. and then use the pairplots.

[@bakkiyarajganesan3130]

I got an idea for Machine Learning algorithm selection. l blindly go with XG boost for nonlinear and imbalance dataset, will get decent results. Your videos are helpful for us and keep doing it 🙏

[@manikantabandla3923]

Does all the classification problems solved using neural networks?

[@connectsantech]

Very nicely elaborated !!! Thanks.

[@anujasebastian8034]

Great insight!!! Thanks Krish!

[@kunikakhobragade6953]

great sir ... can u plz make a video on the comparison of every ML algorithm

[@folashadeolaitan6222]

Thank you for this great insight!

[@karldtrumpeter]

Thanks so much Krish. This knowledge is pricelss.

[@jigarshah1883]

We should give a try to SVN as well as using kernel we can understand the data after plotting and predict it nicely..

[@vineethp8925]

Superb Explanation can you please do on regression

[@RaushanKumar-qb3de]

Thanks. I'm in need of this video

[@ainju1]

Informative. Thank you so much

[@Sunnysinghbhatia]

This was very much helpful Thank you Krish

[@anupambarasia7865]

very well explained thankyou so much sir

[@Amansingh-tr1cf]

U made my day Krish sir

[@Tjos4155]

classic video... cleared my concept

[@gangaholi8600]

Very nice presentation.. thank you..

[@AbhishekBalsara]

Great video, really cleared the concept 👏👏

[@neon91111]

great video man .. luved it..

[@aksharasiri5195]

It's very helpful . Thank you soo much❤😊

[@salihsarii]

This is amazing. Thank you Krish :)

[@MarioLaudazi]

Thanks for the useful videos you post.

[@syedsuhailahmed2868]

Good Explaination.But logistic regression is not linear its segmoid or S shaped

[@samlo4166]

Awesome Video! Thank you for sharing this. I have one question. When there is a lot of overlap, can I use SVM to solve this? thank you

[@minime8745]

Shouldn’t we scale the features before plotting?

[@imteyazahmad9616]

Sir please make more video related to classification and regression .I have to much confusion in regression also,when to select which algorithm .

[@PawanSingh-pf2bl]

Nice video. But I like the way you are presenting at whiteboard. It feels like we are in actual class.

[@prernamistri]

wonderful video. thank you.

[@pratiknaikwade95]

Very useful video krish sir🥰🥰

[@mandarchincholkar5955]

In this case, non linear svm is much better

[@girish-kr]

Awesome Krish Sir

[@syedtasleem2827]

Sir, What about SVM U didn't mention on what kind of Data it can be used after understanding pairplot,..