Add `ml-regression` sample

[maneetgoyal]

Lib. to be used:

1. https://github.com/mljs/regression
2. https://github.com/mljs/random-forest

• All the regression models available in the above 2 packages should be used for training
• All the models should be compared
• Hyper parameter optimization should be performed where applicable
• https://github.com/mljs/param-grid may be useful in hyper param optimization

[qur786]

Can we connect @maneetgoyal sir, on what implementation I have to make.

[maneetgoyal]

@qur786 Try to study this library in the meantime, will ping in sometime

[maneetgoyal]

Code should be added in src/models/regression/index.ts

[maneetgoyal]

https://hackernoon.com/machine-learning-with-javascript-part-1-9b97f3ed4fe5

[maneetgoyal]

Ref: https://people.duke.edu/~rnau/compare.htm

How to compare regression models' performance?

[maneetgoyal]

Sample data set to be used:

1. https://archive.ics.uci.edu/ml/datasets/wine+quality (May be more suited for classification as compared to regression)

[maneetgoyal]

@qur786 Due to missing types, discard, ml-regression. Use the following instead:

• [ ] ~ml-regression-simple-linear~, @maneetgoyal sir (Why this library, we can't use it for multi-independent variable)
• [ ] ~ml-regression-exponential~
• [ ] ~ml-regression-power~🛑
• [ ] ml-regression-multivariate-linear
• [ ] ~ml-regression-theil-sen~ 🛑
• [ ] ~ml-regression-robust-polynomial~

[maneetgoyal]

ml-regression-simple-linear, @maneetgoyal sir (Why this library, we can't use it for multi-independent variable)

@qur786 right, it can be removed

[qur786]

similar to the first one @maneetgoyal sir, ml-regression-exponential and ml-regression-power these two algorithm is similar to the Simple Linear Regression Algorithm.

[maneetgoyal]

@qur786 The regression models deps are added via https://github.com/manufac-analytics/perry/commit/c937ff0ec50498a0e014d3fd60ffe792e6725a72.

will remove ml-regression-exponential shortly

[maneetgoyal]

ml-param-grid didn't have types so https://www.npmjs.com/package/kombi is installed instead. The functionality seems exact

[qur786]

@maneetgoyal sir, changing data from csv to json was a good idea , but most of the dataset that are available for machine learning are of type csv.

[qur786]

@maneetgoyal sir can we use this package from mljs for matrix manipulation if needed?

[qur786]

@maneetgoyal sir , I think this ml-regression-robust-polynomial also similar to ml-regression-robust-polynomial. So I think we shouldn't be working with this as well.

[maneetgoyal]

@maneetgoyal sir, changing data from csv to json was a good idea , but most of the dataset that are available for machine learning are of type csv.

don't worry about, the task in this issue is to do sample modeling; ignore file handiling for now

[maneetgoyal]

@maneetgoyal sir can we use this package from mljs for matrix manipulation if needed?

question unclear

[maneetgoyal]

@maneetgoyal sir , I think this ml-regression-robust-polynomial also similar to ml-regression-robust-polynomial. So I think we shouldn't be working with this as well.

question unclear

[qur786]

@maneetgoyal sir can we use this package from mljs for matrix manipulation if needed?

question unclear

Sorry @maneetgoyal sir, I actually didn't attach the link of the package ml-matrix https://github.com/mljs/matrix , this is a dependency of RandomForest Package as well. So that we can use it to compare, add substract large matrices and this kind of works.

[qur786]

@maneetgoyal sir , I think this ml-regression-robust-polynomial also similar to ml-regression-robust-polynomial. So I think we shouldn't be working with this as well.

question unclear

@maneetgoyal sir as ml-regression-robust-polynomial , the ml-regression-robust-polynomial package also deals with only one independent variable. just like Simple Linear Regression. Should we use it ?

[maneetgoyal]

@maneetgoyal sir can we use this package from mljs for matrix manipulation if needed?

question unclear

Sorry @maneetgoyal sir, I actually didn't attach the link of the package ml-matrix https://github.com/mljs/matrix , this is a dependency of RandomForest Package as well. So that we can use it to compare, add substract large matrices and this kind of works.

yeah it can be use if there is a need

[maneetgoyal]

@maneetgoyal sir as ml-regression-robust-polynomial , the ml-regression-robust-polynomial package also deals with only one independent variable. just like Simple Linear Regression. Should we use it ?

yeah can be skipped

[maneetgoyal]

Fixed via #82