RoboticsClubIITJ / ML-DL-implementation

An implementation of ML and DL algorithms from scratch in python using nothing but NumPy and Matplotlib.
BSD 3-Clause "New" or "Revised" License
48 stars 69 forks source link

Added PCA #101

Closed utkarsh0702 closed 3 years ago

utkarsh0702 commented 3 years ago

Implemented #52

Implemented Principal Component Analysis using NumPy

Other comments

rohansingh9001 commented 3 years ago

@utkarsh0702 there are some build errors in your code which corresponds to wrong formatting which we require. Please use tools like flake8 to correct the formatting or refer to this error log here.

kwanit1142 commented 3 years ago

@utkarsh0702 apart from format errors mentioned by @rohansingh9001 , Why you used scipy library? Also, provide example to show the working of PCA, alongside input dataset. For reference, see other model implementation examples in this Repository

utkarsh0702 commented 3 years ago

@rohansingh9001 I will go through the errors. @kwanit1142 Ok, I will do so. I have used scipy for the gammaln function.

kwanit1142 commented 3 years ago

@utkarsh0702 avoid using libraries other than Matplotlib and Numpy . Instead of using Scipy, try to make utility file of the function required.

utkarsh0702 commented 3 years ago

@kwanit1142 I will try doing so

utkarsh0702 commented 3 years ago

@kwanit1142 Can we add 3 more loss functions: Mean Bias Error, Hinge Loss and Cross-Entropy Loss? If possible can I work on them?

kwanit1142 commented 3 years ago

@utkarsh0702 this implementation is now better than before, but there is one issue u would have to solve, i.e Instead of importing pandas, try to change file type, like in txt format or so, such that numpy can easily fetch it, Otherwise LGTM...

utkarsh0702 commented 3 years ago

@kwanit1142 I will see to it.

utkarsh0702 commented 3 years ago

@kwanit1142 Is it fine? I wanted to ask, how many points a person get for resolving an issue in NWoC?

kwanit1142 commented 3 years ago

@utkarsh0702 regarding point-system queries, @rohansingh9001 can clarify to you. In addition to this, he can review this pull request too. Otherwise, LGTM for now.

utkarsh0702 commented 3 years ago

@rohansingh9001 Please review the pr. Please tell me how many points are allotted to a person for resolving an issue in NWoC?

utkarsh0702 commented 3 years ago

@kwanit1142 Can we add 3 more loss functions: Mean Bias Error, Hinge Loss and Cross-Entropy Loss? If possible can I work on them?

kwanit1142 commented 3 years ago

Looks good to Me Now..