Family Expenditure and Income Analysis using Machine Learning algorithms

Abstract

Expenditure analysis should be done by every household to manage all the expenses of the family. Income prediction gives an overview of the income earned by the household to manage all the family’s financial needs. Based on the expenses and other required data from the user, the system will predict the user’s annual income to meet the expenses. The predicted annual income can be used by the government for initiating policies for the poor people. This prediction task is performed using Decision Tree and Random Forest Regression algorithms, as the data used for this model is continuous. Our proposed Random Forest model predicts with an accuracy of 74.35%. Based on accuracy metrics, our model is compared with Decision Tree, accuracy of 48%. Clearly, the our proposed model is more suitable for classifying than the Decision Tree model. As decision trees are best suitable for predictions based on non-linear data, we cannot depend on a single decision tree for the prediction of income. Bagging technique-based Random forest Regression is made use for the prediction of the income.