TwoVariableRelation

TwoVariableRelation Package is intended to establish the relationship between two features of the dataset.It starts with performing multivariate outlier detection on those two features of a dataset to remove outliers based on density and distance.Further,it divides the dataset into train-test(70:30), fits polynomial models on training data up to polynomial degree based on input given by the user.Capturing all inputs, it recommends the best model based on train RMSE and test accuracy.

Installation

Dependencies

TwoVariableRelation Package requires:

• Numpy
• Pandas
• Scikit-learn
• Scipy

Install these libraries before proceeding further.

User Installation

Use the package manager pip to install TwoVariableRelation.

pip install TwoVariableRelation

Usage

import TwoVariableRelation

# Returns table with metrices(RMSE,Rsquare and Accuracy) for polynomial model from degree (0-N) and recommends the best model.

TwoVariableRelation.polynomial_model(N,X,Y)

where:
N=Maximum degree to consider for training a polynomial model.
X=Pandas series of Independent Variable.
Y=Pandas series of Dependent Variable.

Result

Following is the result generated when package is called with below mentioned argument on the dataset:

TwoVariableRelation.polynomial_model(3,X=df.iloc[:,0],Y=df.iloc[:,1])

Order	Train RMSE	Test RMSE	Train Rsquare	Test Rsquare	Test Accuracy
0	9.829	10.534	0.000	-0.179	75.595%
1	9.489	9.103	0.068	0.120	74.737%
2	9.463	8.627	0.073	0.210	76.590%
3	9.463	8.640	0.073	0.207	76.607%

Model is best fitted at polynomial degree 3

*Accuracy is calculated with following formula:

Accuracy=[1-[abs[(sum of actual Y_test)-(sum of predicted Y_test)]/Sum of actual Y_test]]*100

License

MIT