FACTORS AFFECTING THE VARIATION OF TECHNOLOGY STOCKS USING BY DATA MINING

Abstract

The objectives of this study were to study factors affecting to variation of the monthly closing price of Technology share and to compare efficiency of model by data mining technic such as Decision Tree (J48), Random Forest, K- Nearest Neighbor: KNN and Neural Network in order to find the right forecasting model for the monthly closing price of Technology share and to study factors affecting to variation of the monthly closing price of Technology share by multiple regression analysis. Data used were collected from The Securities Exchange of Thailand for the past 5 years from 2013-2017, the total number of shares was 36 shares. The factors that affected the variation were the following 10 elements; Close, Market Cap, P/E, P/BV, Book Value Per Share (BVPS), Dividend Yield   (DivY), Turnover Ratio, Return On Equity (ROE) , Return on Assets (ROA) and Tota Asset. The data was analyzed by Multiple Linear Regression Analysis. It was found that variation of the Technology share which affected to the change in closing price of each month, the most was Book Value Per Share (BVPS), secondary was Return on Assets (ROA) and Return On Equity (ROE), respectively. The model test was found that classification by K- Nearest Neighbor: KNN had been the most effective, the accuracy value was 100% and the average deviation was 0.00% following by Random Forest, the accuracy value was 100% and the average deviation was 0.14 % besides Neural Network, the accuracy value was 81.11% and the average deviation was 0.24 and Decision Tree (J48), the accuracy value was 76.66% and the average deviation was 0.35%, respectively as efficiency. The result of this study has benefit for investors who will invest in Technology share must aware of Book Value Per Share (BVPS), Return on Assets (ROA) and Return On Equity (ROE), respectively.