Live Virtual Interview For Data Science Fresher By Krish And Sudhanshu

Please donate if you want to support the channel through GPay UPID,

Please join as a member in my channel to get additional benefits like materials in Data Science, live streaming for Members and many more

Please do subscribe my other channel too

Connect with me here:

[kshitizomar]

Although it was his interview, here while watching, I am sweating lol. Great session!

[comatosetorpor]

he is in 3rd year and he has a lot to learn, but certainly on the right track and with the right pace.

[karanyadav-fzju]

Though his few answers were wrong but his confidence was on top, gotta appreciate him!

[arunmohan]

* skweness measures how much your data is distorted.Right skweness means positive skweness and the mean will be greater than the median. To remove skweness u can apply log over it to normalize.

* Standard scalar normalize the data as x-mean/STDdev. While min max scalar make it betwee 0 to 1.

* gini impurity = 1 - sum(p(y)^2)

* n_estimators=30 can reduce variance more..since rf generally reduce variance...so there is a chance that it can underfit...Generally it depends on data.


* Machine learning over deep learning when we have less data and interpretability matters for our model.

* Activation functions like relu, sigmoid etc introduces non linearity to the network.

* We can SGD with decay learning rate if we want to reduce it over iteration

* Another reason for up and down movement of loss can also be due to exploding gradient problem..

[engineerjunaid]

After watching this interview I am feeling nothing I have to revise lots of things Very good effort by the organizers and by the Manish

[shreyasb.s]

Really awesome interview. Thanks Krishna and Sudhashanu for wonderful questions. Its helps lot for me. Its clearly says even myself need to improve

[vikaschavan]

After watching this interview I realised I need to do more hard work and much practice..

[indrapatel]

Even though he is a fresher, he is performing excellent with confidet with out tense or nervous really very good

[poojamangal]

Manish is really great guy and now he is so improved. Great to see this.

[BHRIGUCHADHA]

Great effort, Krish. You are helping a lot of intellectuals.

[vishalvenkata]

Got to know how complicated the interview at the same tym got the structure of interviewing through this mock interview...good one!

[umeshkaushik]

One thing is he was quick in answering those unpredicted questions, I would have been sitting like dumb and thinking a lot.

[Gagan]

Heads off to his confidence even after complex questions

[satyendrasingh]

A Great Reality check for fresher, Thanks for sharing!

[hashiharis]

The interview questions are good and even the guy who is being interviewed is performing with confidence. But what I have a suggestion is, even though this interview had good questions, the way of conducting a professional interview should not be like this. An interview is not meant to stress the person who is being interviewed and confuse him. It is the responsibility of the interviewers to conduct an interview in the most peaceful calm way. We are humans and mistakes would be answered by everyone, anyone can be wrong, but the mistakes should not be pointed out in the middle of the interview, which inturn will drain the confidence of the interviewed guy. This kind of attitude is not expected from a professional interviewer. It becomes the responsibility of the interviewer too that if they point out a mistake they should correct the mistake by letting them know the concept or the right answer. Honestly, interviews are meant not to stress or pressurise the person or the people watching this mock interview. So it would be better to conduct such interviews in a modern way. Research the interview rules and post mock interview videos instead of pressurising the freshers. They do deserve a respect for their hardwork from the interviewers, even if they are not getting job atleast behave to them with a consideration as they are humans and not as a machine learning model trained with some stress parameters.

[newbie]

25:51, tanh is more computationally intensive than sigmoid, see the formula for both functions, you'll notice that the exponential term appears twice in tanh, making it more difficult to computer.

Also sigmoid are not zero centered, this makes convergence slower
Sigmoids and tanh functions also saturate often, making the gradient of the activation==0, which stops the learning process for neurons

[sharathraghuveer]

I just realized that I need to work even more harder in interview aspect :0

[riyabiswas]

We really wish for more interviews like this ....
Thank you very much

[MrJjthecronfan]

This is pure gold @krish @Sudhannshu great questions and beautiful knowledge drilling thanks a ton . @manish nice confidence and effort.

[durgeshgupta]

Very informative video 👍 Thanks for sharing.