Selection and Forecasting of Stock Market Patterns using K-mean Clustering

Stock Markets being dynamic entities make the stock price prediction a challenging task. A number of models and methods are in vogue for predicting the stock prices [5,6,7,8]. In this paper we have used silhouette function to select the best cluster amongst the clusters obtained by K-mean clustering of the stock market data.Then ARIMA model is applied to forecast the stock prices. By application of this method, the percentage error is decreased by at least 2% when compared to forecasting done by without clustering.