Analysis of Clustering-Based Stock Market Prediction

Abstract

Estimation of the economy of a developed country is affected by the performance of share market of the country. Predicting stock prices has been a major research challenge. There is an influx of various approaches that attempt to address this problem. Stock markets generate massive data, almost as much as social media, which can be used to identify current trends, accordingly price of stocks can be predicted thereby yielding profits to those invested. Existing stock market models are complex. Stock prices are volatile in nature, the determination of which is difficult to estimate as it is influenced by numerous parameters even small news can prompt an increase or decrease in price. Clustering refers to grou** of similar objects in order to form a cluster. Many surveys have been done on clustering that shows the worthy enactment on datasets for cluster formation such as K-means and fuzzy c-means. Similarly, the use of machine learning algorithms can test and train the data to find out the best possible way to forecast the values of data and predict where the stock prices will move toward. We conducted a study and found that before applying machine learning algorithms we must apply clustering to the dataset to make it tailor ready for the predictive algorithms. This paper surveys different approaches of clustering and presents a better-formed way to apply clustering to stock prices.