ABSTRACT
Economists have conducted research on several empirical phenomena regarding the behaviour and neural activities of individual investors, such as how their emotions and opinions influence their decisions. All those emotions and opinions are described by the word Sentiment. In finance, stochastic changes might occur according to investors’ sentiment levels. This chapter represents the mutual effects between some financial process and investors’ sentiment with Multivariate Adaptive Regression Splines (MARS) model. Furthermore, we consider to extend this model by using distinct data mining techniques and compare the gain in accuracy and computational time with its strong alternatives applied in the analyses of the financial data. Machine Learning methods are well-known and beneficial tools for prediction problems and have already been successfully applied to numerous financial associated problems. In this study, apart from pure financial related problems, we focus on behaviour of financial problems which is based on the investors' behaviour introduced as Sentiment. The goal of this study is to compare the forecasting performance of sentiment index by using single MARS, Random Forest (RF), Neural Network (NN) models, and two-stage MARS-NN, MARS-RF, RF-MARS, RF-NN, NN-MARS, and NN-RF models. principal component analysis for association tests and smoothed functional principal component analysis for association tests are presented. Operations research and machine learning can make use of each other. One of the important advantages is that this relation between them raises the applications of operational research, especially, in data science through the employ of machine learning.
