Predictive analysis on stock market data

Abstract

In today’s digital world data generated is huge and heterogeneous in nature. This data must be analyzed and organized for efficient usage. Predicting unseen future events by analysis is a methodology of progressive analytics. This division uses many modes of operandi staring from statistical modelling, excavating data using data mining, learning and gaining knowledge using machine learning and artificial intelligence. This work presents predictive analytics on stock market data. Predicting stock market performance is the most difficult things to do. Stock prices move up and down every minute due to fluctuations in supply and demand. The altering stock prices makes it arduous for accurate prediction. The predicted values are of immense use to the stock investors. This enables the investors to take right decisions during trading. Investing in right time at an appropriate area is crucial. Stock market prediction accomplishes this task effortlessly. In this paper, we are using data mining techniques. After detailed study on the possible algorithms, k-means was the most suitable clustering algorithm. K-means allows faster computation and produce tighter clusters. To get the finest close results, classification technique is also implemented along with clustering. Decision tree classifier is used to build a model. The upcoming stock values up to five days are predicted. Results are tested on the SBI stock data set which is collected from the national stock exchange limited. A comparative study is conducted using SVM classifier to prove that methodology using decision tree classifier is best suited for predicting the result.