Automated Data Preprocessing and Training Interface for Machine Learning Applications

Abstract

Machine learning is a promising tool for analyzing and predicting data. However, it is very time-consuming and challenging to manually process the data. There are also many missing values in the data that can affect the accuracy of the models. We propose an automated approach to pre-process the data. It efficiently performs the various steps involved in the data cleaning process by converting the categorical values to a label encoded value, sampling the data, and replacing the missing values with the appropriate central tendency. This project takes the raw data as input and provides a clean error-free dataset that is suitable for training machine learning models. It also aims to use a web application to automate data pre-processing and training of classification and regression models. It uses multiple algorithms and normalization strategies to get the optimum model based on the desired metric. It allows users to download the clean dataset as well as the trained models after the models have been trained. Multiple datasets, including binary classification and regression, are used to test this model.