High bias: Parametric models Low bias: Non-parametric models. Thanks for the video!

Thanks for this video, I really appreciate it.

Hi Aman, very nice explanations. please find the below answers. The parametric models has high bias due to simplified assumptions on the data(i.e. data is linearly separable).Because of high bias we may have underfitted models which high training error and high CV error . The non-parametric models are overfitted models to the input data. They have low training error and high CV error. when there is any change in the training data the training error also increases.

I was exactly looking for explanation on this topic and your video answered all my questions! Again, Thank you for your great work!

@tempura_edward4330

3 жыл бұрын

So parametric models tend to have more bias and non parametric models tend to have less bias but more variance.

@UnfoldDataScience

3 жыл бұрын

Thank you. Yes. Right answer.

The basic idea behind the parametric method is that there is a set of fixed parameters that uses to determine a probability model In non parametric model, there is no fixed set of parameters available, and also there is no distribution (normal distribution, etc.) of any kind is available for use. This is also the reason that nonparametric methods have high accuracy. Therefore A non-parametric model will always have a higher prediction accuracy compared to a parametric model.

@UnfoldDataScience

2 жыл бұрын

Yes true, Sanyam.

loved it. thanks

Very well explained!

hi Aman, very clear explanation, appreciate the effort. Could you please help on statistical parametric and non parametric tests, when to use parametric and when to use non parametric tests

Great explanation thank you Q How is Gaussian process regression non-parametric, I mean it assumes something at first which is the kernel. if we are assuming a prior how can we say something is non-parametric. Can you please explain this

in non parametric there should be low bias due to overfitting and in parametric there should be high bias cause of underfitting.

thanks Aman, very clear explanation

@UnfoldDataScience

2 жыл бұрын

My pleasure Assad.

Thank you so much sir ❤

Great

@UnfoldDataScience

6 ай бұрын

Thank you

Thanks

thank you... cleared ans is non parametric group will have low bias as the work on population data

Thanks, Bro More Videos like this

@UnfoldDataScience

3 жыл бұрын

Hi Ajay.

Hi Sir.Very Crystal Clear..Superly Explained..When Can we Expect Another Mock Interview get Uploaded..Thank you..

@UnfoldDataScience

3 жыл бұрын

Thanks Kirandeep.

sir , how are all these implemented in real life . could you please explain?

we prefer non parametric models over parametric models for solving our problems. correct me if i am wrong?

thanks Sir, nicely explained

@UnfoldDataScience

2 жыл бұрын

Welcome Someshwar.

😍

How the new data is handled after the model is moved to production. Example: During model development the categorical data is converted to 1 and 0 using one hot encoding... When the new data is applied in production how the categorical data or text data is processed..

@UnfoldDataScience

3 жыл бұрын

Very good question, all the preprocessing should happen on new data as well.

Very good content.

@UnfoldDataScience

3 жыл бұрын

Much appreciated

finished watching

great video!!

@UnfoldDataScience

Жыл бұрын

Thank you!!

thank you so much

@UnfoldDataScience

2 жыл бұрын

Welcome.

Also - when you say we need more data for non parametric, could you explain how much data is needed please

@UnfoldDataScience

3 жыл бұрын

Depends, at least 50k I would say.

Thank you!!!!

@UnfoldDataScience

Жыл бұрын

You're welcome!

Thank you- could you please do non parametric regression in Python? Thank you

@UnfoldDataScience

3 жыл бұрын

Will try to upload.

Good but specking speed need to must me increase

Nice video sir

@UnfoldDataScience

2 жыл бұрын

Thank you

great explanation

@UnfoldDataScience

3 жыл бұрын

Glad you liked it

Nice explanation bro

@UnfoldDataScience

3 жыл бұрын

Thank you 🙂

@sudheeshe1384

3 жыл бұрын

@@UnfoldDataScience bro please do the video on L1 & L2 regularization

Collar niche rehta to ek decent teacher wali feeling aati video dekhne me. Bt aisa laga as if apna sutta partner samjha raha ho kch technical baatey.

Non parametric- low bias Parametric - high bias

@UnfoldDataScience

Жыл бұрын

Yes

Non peremetric data becz giving high data

Based upon the explanation, I will say, as parametric learning algorithms are provinding low fit models, they will have 'high bias'. As a result, they will perform poor (if compared with non-parametric ML algos) on both train and test data. On the other hand, as non-parametric algorithms tends to overfit, they might perform well with train data, but on real life data performace may degrade. So this is a case of 'high variance'. But I have a small doubt, when you said, we assume something about f(x) [and you gave a very nice real world example], what assumptions were you trying to imply? (I mean in terms of dataset, what are those assumtions, that we make on dependent variable of our dataset)

@UnfoldDataScience

3 жыл бұрын

example like "Salary" is linear function of "experience".

High Bias: Parametric? Low Bias: Non parametric?

Bhai thoda 2x mai bola karo, subeh exam dene b jana hai

Thank you soooo much 🤍🤍✨, i was afraid from my final exam but now I’m not 😌

@UnfoldDataScience

Жыл бұрын

You're welcome 😊

@brianocorner6467

Жыл бұрын

Pass HOA 😂😂??

loved it. thanks