Data Mining for Knowledge Discovery from Financial Institution Database

Abstract

Due to its importance for the investment decision-making and risk management, describing and predicting stock represents a key topic in stock analysis. Stock or shares are valued and analyzed by stock investors using fundamental analysis and technical analysis. Stock investors describe and predict stock manually based on individual experiences. Employing manual procedure in analyzing stock, most especially technical analysis is always very cumbersome and inefficient; because of much time that is consumed in examining the past records of a respective firm, an investor is willing to invest in. This paper presents a better way of describing and predicting stock, especially technical analysis, by employing data mining techniques. A database was developed employing 360 records of daily activity summary (equities) and 78 records of weekly activity summary (equities) spanning through 18 months that is, from January 2007 to June 2008. These data were obtained from the daily official list of the prices of all shares traded on the stock exchange published by the Nigerian Stock Exchange being the financial institution that runs Nigerian stock market; using banking sector of Nigerian economy with three banks namely:- First Bank of Nigeria Plc, Zenith Bank Plc, and Skye Bank Plc. A data mining software tool was developed and employed in identifying patterns and relationships from the database to generate new knowledge about the data set in the database through the use of data mining techniques that employ regression analysis.