Saving Patterns and Investment Preferences: Prediction Through Machine Learning Approaches

Abstract

Capital formation is the essence of economic development for a developing economy like India. However, it needs continuous efforts by the investors as they invest in different financial instruments. Recently individual investors are becoming more vigilant while selecting these instruments and making an investment, which makes it difficult for the financial service providers to provide the product and services as per the need and demand of these individual investors. To address this issue the financial companies should study the financial behavior of the individual investors, they should know what factors are affecting the saving and investment patterns of the investors. In continuation to this, the current study is conducted in the region Uttarakhand. In this study first of all machine learning tools (KNN, Tree, SVM, Random Forest, Neural Network. Linear Regression, AdaBoost) were run on the data set to find out the best machine learning tool. In the second step, the best machine tool is used statistically. From the regression results, it can be inferred that the saving patterns of the small investors can be predicted by the eight variables—age, gender, qualification, occupation, annual income, marital status, dependents in the family, and geographical area. However, investment preferences are related to the qualification, occupation, annual income, marital status, and dependents in the family. They can be used as a predictor to forecast investment preferences. Thus, the financial service companies can predict the investment behavior (saving patterns and investment preferences) of their customers and make their marketing strategies based on the regression equation constructed.