6.4. How to choose the right Machine Learning Model | Model Selection | Cross Validation

[@manojchoukiker3232]

Very nice explanation i always confuse thanks sir for differentiate and clearly explanation with example please keel continue.....

[@Siddhardhan]

Always welcome😇

[@ErLakhan]

I was also looking for same, Thank you

[@sachinvithubone4278]

This is good explanation. Thanks. Can we get videos on overfitting and underfitting.

[@Siddhardhan]

you're welcome 😇 those videos will be posted tomorrow and Friday.

[@sachinvithubone4278]

@@Siddhardhan thanks for response 🙏

[@liftsbysuku]

Thank you If I crack any big MNC you are one of the reason

[@Siddhardhan]

that would be a great news!😇 all the best!

[@YoutubeVoyageur]

For text, RNN are not very interesting (transformer or other autoencoders versions are better). RNN (I include GRU/LSTM) should be better on "numerical" time series. For numerical data (strange term as all data in ML are "numerical" as in the input of models) I think that SVM, decision trees and logistic regression are almost always under optimized by comparison with MLP and Gradient Boosting / Random Forest that are simply the same thing but bigger and smarter. Thus I don't understand the link between cross validation (that is a validation method particularly usefull with few data) and the model selection. By the way good luck for the following, it's a good initiative.

[@Siddhardhan]

hi! Gated Recurrent Unit (GRU) and Long Short Term Memory(LSTM) are like the evolution of RNN. They were developed to overcome the disadvantages of RNN. and in the video, I mentioned that I don't want to go into the details in this particular video as it's a topic to be discussed in an advanced stage. About numerical data, I didn't mentioned that it's based on the input to the models. I meant the raw data that we have.

[@sandipansarkar9211]

finished watching

[@ranganayakuluchabala7342]

Sir, can you give the link to the material for the complete machine learning

[@Siddhardhan]

hi! I cannot share the ppt

[@tendulkartejesh3318]

@@Siddhardhan But in the video description, you have mentioned that, you are providing the ppt as pdf for Rs.200. What about that bro? If yes, then please send me bro, I'm ready to pay for that bro. I need it for revising the contents daily bro.

[@mdarifhassan9064]

hi bro . what is the best model for identity farud (USER BEHAVIOUR Authentication) dection in AI/ ML/DL?

[@mjanish9836]

SVM

[@tanishqrastogi1011]

classification models but mainly logistic and svm in this case would be the better models

[@VivekKumar-rn5dc]

Plz Make a Deep Learning Project using CNN algorithm .

[@Siddhardhan]

will do in deep learning course

[@universal4334]

What if outlier is there in cross validation

[@Siddhardhan]

in that case, we may need to remove the outliers during data Pre-Processing part before the cross validation.

[@gauravfamily2209]

Pls discuss also type of cross validation. It is important part of ML.

[@KumR]

you keep saying model. guess u mean Algo?

[@raghavsinghal22]

I'm waiting for algorithms and it's intuition

[@Siddhardhan]

very soon!

[@raghavsinghal22]

@@Siddhardhan Can we learn algorithms in random manner or we have to follow sequence?

[@subzero4787]

Can u make this video in hindi

[@Siddhardhan]

I don't know Hindi