[Paper Review] Turnover Prediction Of Shares using Data Mining techniques : A Case Study

D. S. Shashaank, V Sruthi+2 more|arXiv (Cornell University)|Aug 1, 2015

Stock Market Forecasting Methods3 references3 citations

TL;DR

This study proposes a data mining approach to predict daily stock turnover using historical market data, employing feature selection via Boruta and classification models (Random Forest, SVM, Decision Tree, Multinomial Regression) in R. It achieves a 95% prediction accuracy, demonstrating high effectiveness in forecasting turnover and identifying key market attributes influencing stock trading volume.

ABSTRACT

Predicting the turnover of a company in the ever fluctuating Stock market has always proved to be a precarious situation and most certainly a difficult task in hand. Data mining is a well-known sphere of Computer Science that aims on extracting meaningful information from large databases. However, despite the existence of many algorithms for the purpose of predicting the future trends, their efficiency is questionable as their predictions suffer from a high error rate. The objective of this paper is to investigate various classification algorithms to predict the turnover of different companies based on the Stock price. The authorized dataset for predicting the turnover was taken from www.bsc.com and included the stock market values of various companies over the past 10 years. The algorithms were investigated using the "R" tool. The feature selection algorithm, Boruta, was run on this dataset to extract the important and influential features for classification. With these extracted features, the Total Turnover of the company was predicted using various classification algorithms like Random Forest, Decision Tree, SVM and Multinomial Regression. This prediction mechanism was implemented to predict the turnover of a company on an everyday basis and hence could help navigate through dubious stock market trades. An accuracy rate of 95% was achieved by the above prediction process. Moreover, the importance of stock market attributes was established as well.

Motivation & Objective

• To address the challenge of predicting stock turnover in a volatile market, where traditional methods suffer from high error rates.
• To evaluate the performance of multiple classification algorithms in forecasting daily turnover based on historical stock price data.
• To identify the most influential market attributes affecting turnover through rigorous feature selection.
• To develop a practical, daily forecasting system to guide investment decisions in uncertain stock markets.
• To achieve high prediction accuracy using data mining techniques on a decade-long dataset of company stock values.

Proposed method

• The study uses a 10-year dataset of stock market values from a public source to train and test prediction models.
• Boruta, a feature selection algorithm, was applied to identify the most relevant and influential features from the dataset.
• The selected features were used to train and evaluate four classification algorithms: Random Forest, Decision Tree, SVM, and Multinomial Regression.
• All models were implemented and compared using the R programming language for statistical computing and data analysis.
• The prediction task was framed as a classification problem to forecast daily turnover levels with high precision.
• Model performance was evaluated based on accuracy, with the best-performing model achieving 95% accuracy.

Experimental results

Research questions

• RQ1 Which classification algorithms perform best in predicting daily stock turnover using historical market data?
• RQ2 Which market attributes are most influential in determining stock turnover according to the data?
• RQ3 Can feature selection using Boruta improve the accuracy of turnover prediction models?
• RQ4 To what extent can data mining techniques reduce prediction error in stock turnover forecasting?
• RQ5 Can a daily turnover prediction system be effectively built using standard machine learning tools?

Key findings

• The proposed method achieved a prediction accuracy of 95%, indicating strong performance in forecasting daily stock turnover.
• Random Forest, SVM, Decision Tree, and Multinomial Regression all demonstrated high predictive capability, with the best model reaching 95% accuracy.
• The Boruta algorithm successfully identified the most relevant features, enhancing model interpretability and performance.
• Key market attributes such as stock price, trading volume, and market capitalization were found to be highly influential in turnover prediction.
• The study confirms that data mining techniques can significantly reduce prediction error in financial forecasting tasks.
• The framework enables daily turnover predictions, offering practical utility for navigating uncertain stock market conditions.