Impact of Training and Testing Data Splits on Accuracy of Time Series Forecasting in Machine Learning

Abstract

Machine learning is the process of developing Artificial Intelligence in computers. Machine learning has the ability to make a computer perform some task without actually programming it and with minimal human efforts. Machine learning models are trained using appropriate learning algorithm and training data. Here the data is divided into two parts; Training and Testing data. The model will learn to perform a task using the training data and testing data is used to verify if the model works correctly. In this work a Machine learning model predicts weather parameters using Gaussian Process with RBF kernel. The basic aim is to analyse as to how the accuracy of prediction will vary with the different combinations of training and testing data. For this purpose experimentation is carried out with different combinations of training and testing data. The Mean Absolute Error is calculated by comparing the actual values from testing data set and predicted values from the model.