Stock price prediction with machine learning methods

In the financial area, many studies have been performed to estimate the stock prices, which are affected by numerous parameters. In order to contribute to the ongoing research and studies in the financial area, the closing price of shares in different sectors will be estimated by using financial data and machine learning algorithms. At the end of the study, financial variables and machine learning methods with high impact on the closing price will be determined for the selected shares. At the same time, the estimation result for the shares in different sectors will be compared and sectoral differences will be analyzed. Stock shares with the highest average market value in 2016-2018 will be determined and grouped separately and sector-based examinations will be analyzed for the first fifteen shares. boosted decision tree regression, neural network regression, linear regression, decision forest regression, poisson regression, bayesian linear regression will be used in the machine learning algorithms for estimating financial data. For the period between 01.01.2016 and 31.12.2018, daily and stock based transaction and price information, gold price, foreign Exchange price, bond index interest information and daily exchange values of all variables are determined as financial variables in the data set. For the first fifteen shares with the highest market value in the selected period and for all shares traded on Borsa İstanbul, learning and estimation will be done by creating separate data sets. Furthermore, when the parameters of the algorithms are changed, the effect on the result will be evaluated. At the end of the study, financial variables and machine learning algorithms with high impact on the closing price will be determined for the selected shares. At the same time, the estimation result for the shares in different sectors will be compared and sectoral differences will be evaluated.