pyevals 1.3.8

A simple Python Package for Model Evalutaion

Table of Contents

• About the Project
• Installation
• Usage
• Future Work
• Contact

About The Project

A very elegant and simple library to evaluate models.

This module builds the BarPlot, BoxPlot, CountPlot, DistPlot, HeatMap, PairPlot and ViolinPlot only with one line of code. A folder is created 'Plots' where the pdf files of all the plots are stored. Along with this, a pdf file will be generated 'FinalPlots.pdf' which contains all the plots with which EDA can be performed easily.

This module will evaulate the Classification problems and Regression problems with 12 and 6 algorithms respectively.

The Classification algorithms are KNN,LogisticRegression,DecisionTreeClassifier, RandomForestClassifier, SupportVectorClassifier, QuadraticDiscriminantSnalysis, SGDClassifier, AdaBoost, CalibratedClassifier, MultinomialNB, BernoulliNB, GaussianNB.

The Regression algorithms are LinearRegression, PolynomialRegression, RidgeRegression, LassoRegression, SupportVectorRegressor, GradientBoostingRegression.

We also have implemented the Adjusted R Squared method as the Regression Metric Evaluation.

In Classification , Highest Accuracy is Highlighted in Yellow colour.

In Regression , Least Error is Highlighted in Yellow colour.

Installation

1. Clone the repo

gh repo clone Anand-gokul/pyevals

2. Install using pip or pip3

pip3 install pyevals

(or)

pip install pyevals

Usage

import pyevals

# For Exploratory Data Analysis (or) For building the plots

pyevals.BuildPlots(data,CategoricalFeatures,ContinuousFeatures)

'''CategoricalFeatures and the ContinuousFeatures are the lists of the Categorical
and Continuous Features of the dataset respectively. '''


# For Classification

Object = pyevals.build(x_train,x_test,y_train,y_test,'classification')
Object.evaluate()

# For Regression

Object = pyevals.build(x_train,x_test,y_train,y_test,'regression')
Object.evaluate()

Future Work

In this version we are only providing the reports and the plots as many as possible. We are working on improving the plots for better EDA.We will try to implement hyperparameter optimization techniques to get the better results. We will also try to implement other algorithms in classification and regression soon.

Contact

Githublink - https://github.com/Anand-gokul/pyevals

Sai Gokul Krishna Reddy Talla - @Krish - [e-mail]

Ananda Datta Sai Phanindra Tangirala - @Anand - [e-mail]

Anirudh Palaparthi - @Anirudh - [e-mail]