See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/277409163

Stock Trend Prediction Using Regression Analysis – A Data Mining Approach

Article · January 2011

CITATIONS READS

28 6,836

3 authors, including:

Adewole K. S.
University of Ilorin
38 PUBLICATIONS   684 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

EEG-based Communication with a Predictive Text Algorithm View project

Anomalies detection in social networks View project

All content following this page was uploaded by Adewole K. S. on 31 May 2015.

The user has requested enhancement of the downloaded file.

 Volume 1 No. 4, JULY 2011 ISSN 2222-9833
ARPN Journal of Systems and Software

©2010-11 AJSS Journal. All rights reserved

http://www.scientific-journals.org

Stock Trend Prediction Using Regression Analysis –

A Data Mining Approach
1
S Abdulsalam Sulaiman Olaniyi, 2 Adewole, Kayode S. , 3 Jimoh, R. G
1 2
Al-Hikmah University,Ilorin, Nigeria, Kwara State University, Malete, Nigeria, 3 University of Ilorin,Ilorin, Nigeria.
1 2 3
[email protected], [email protected], [email protected]

ABSTRACT
Organizations have been collecting data for decades, building massive data warehouses in which to store the data. Even
though this data is available, very few of these organizations have been able to realize the actual value stored in it. The
question these organizations are asking is how to extract meaningful data and uncover patterns and relationship from their
databases. This paper presents a study of regression analysis for use in stock price prediction. 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 using
banking sector of Nigerian economy with three banks namely:- First Bank of Nigeria Plc, Zenith Bank Plc, and Skye Bank
Plc to build a database. A data mining software tool was used to uncover patterns and relationships and also to extract
values of variables from the database to predict the future values of other variables through the use of time series data that
employed moving average method. The tools were found capable technique to describe the trends of stock market prices
and predict the future stock market prices of three banks sampled.

Keywords: Data warehouses, regression analysis, stock price, data mining, moving average

1. BACKGROUND TO THE STUDY investors in making complex decision in stock market

activities thereby reducing the risk in stock investment.
Across a wide variety of fields, different data are
being collected and accumulated at a dramatic pace. The 2. METHOD USED
amount of these data generated and stored is growing
exponentially, due to the continuing advances in computer The data obtained from the daily and weekly
technology. This presents tremendous opportunities for activity summaries (equities) published by Nigerian Stock
those who can unlock the information embedded within Exchange was used in building a database. A data mining
this data. However, these challenges has called for an software tool was developed which employed the use of
urgent need for a new generation of computational regression analysis through the use of time series data that
theories and tools to assist humans in extracting useful employed moving average method to predict future stock
information (knowledge) from the rapidly growing market prices.
volumes of digital data. These theories and tools are the
subject of the emerging field of knowledge discovery in 2.1 Regression Analysis
databases (KDD) [1][2]. Knowledge discovery out of the
collected data can be done using a technique called Data Linear regression is one of the most common data
Mining. A commonly accepted definition of data mining mining technique for predicting the future value of
which is a process of selecting, exploring and modelling variable based on the linear relationship it has with other
large quantities of data to unravel previously unknown variables [5][6][7][8]. Basically, it assumes there is a
patterns for the purpose of business and commercial straight line that approximates the data set, and bases the
advantage [3]. In this paper, the serial movement of stock forecast on it.
prices over a period of time extracted from the daily There is only one independent variable and the
official list of Nigerian Stock Exchange, are used in formula that describes this relationship is the one that
building a database and values of variables were extracted defines a straight line:
from the database to predict the future values of other y  a  bx ------------------------ 1.1
variables through the use of time series data that employed
moving average method. Where, y is the dependent variable
This research also investigates how to make use x is the independent one.
of this rich data to predict financial market prices in the a and b are the line’s coefficients.
banking sector of Nigerian economy using three banks as
a case study. We predict stock market prices using Moreover, the least square method is employed in finding
information contain in daily and weekly activity the regression line.
summaries (equities) published by Nigerian Stock The least squares method is expressed as:
Exchange. The uncovered patterns and the predicted
values from the database assist the stock brokers and y  a  bx ---------------------- 1.2

154
 Volume 1 No. 4, JULY 2011 ISSN 2222-9833
ARPN Journal of Systems and Software

©2010-11 AJSS Journal. All rights reserved

http://www.scientific-journals.org

Where, a  y  bx (P.E) was used for each of the banks. For example, 3-
monthly average can be calculated from the data given by

n xy   x  y Janp.e + Febp.e + Marchp.e , Febp.e + Marchp.e + Aprilp.e,

b= --------------------- 1.3
n x 2   x 
2 3 3

Marchp.e + Aprilp.e + Mayp.e, ........... 1.8

In accordance with this paper, each variable denotation is 3
as follows:
y = current market price and these total to be written beside the month of February,
x = percentage earning (P.E) ratio March, April and so on.
y  Mean occurrence of current market price
3. ANALYSIS OF RESULTS
x  Mean occurrence of percentage earning (P.E.) ratio
n  Total number of occurrence of the variables Data mining is a step in the KDD (Knowledge
Discovery in Database) process that consists of applying
The above variables were derived from the data analysis and discovery algorithm that, under
relationship between the percentage earning (P.E.) ratio, acceptable computational efficiency limitations, produce a
current market price and the earning per share (EPS), particular enumeration of patterns (or models) over the
described by the formula data. The data to be mined is first extracted from an
enterprise data warehouse into a data mining database or
Percentage earning (P.E.) ratio = current market price data mart. This means that for a data mining tool to work
efficiently and effectively, database must be built. In this
earning per share ---1.4
paper, we built a database for the data mining software
tool using the information obtained from the daily activity
The straight line trend in equation 1.1 was obtained for summary (equities) published by Nigerian Stock Exchange
each of the banks as shown below spanning through eighteen months. Data obtained were
analysed and summarized as shown in Figure 1.0. The
First Bank Plc discovered data are used to generate new knowledge about
the data in the database and the identified patterns are
y  35.85  0.12 x ------------ 1.5 relationships are depicted in Figure 1.1, which shows the
Skye Bank Plc market trends of stock prices for the three banks between
y  30.97  0.64 x -------------1.6 January 2007 and June 2008.
Furthermore, we extracted values of variables
Zenith Bank Plc
from the discovered data to predict six months future
y  14.39  1.16 x -----------------1.7 values for the months of July, August, September,
The method of moving average was used to get the October, November, December, year 2008.
corresponding value of x using 3-monthly moving average
period.

2.2 Moving Average

Moving average method is a device for reducing
fluctuations and obtaining trend values with a fair degree
of accuracy. This method consists of taking arithmetic
mean of the values for a certain time span and placing it at
the centre of the time span. In this method the average
value of a number of years (months, weeks or days) is
taken as the trend value for the middle point of the period
of moving averages. The process of averaging smoothens
the curve and reduces the fluctuations [1]. A moving
average is plotted as a line on a graph of price changes.
When price fall below the moving average, they have a
tendency to keep on falling. Conversely, when price rise
above the moving average, they tend to keep on rising [4].
For applying the method of moving averages the period of
the moving averages has to be selected first. In this paper,
3-monthly moving average of percentage earning ratio
Figure 1.0: Generated Data from the database
155
 Volume 1 No. 4, JULY 2011 ISSN 2222-9833
ARPN Journal of Systems and Software

©2010-11 AJSS Journal. All rights reserved

http://www.scientific-journals.org

Figure 1.1: Identified Patterns and Relationships describing

the generated data

Moreover, our system predicts monthly

movement of stock prices for each of the three banks using
regression analysis. The regression coefficient a and b in
equation 1.2 are then solved using the method of least
squares to generate the linear functions in equations 1.5, Figure 1.3: A line graph showing trends of the predicted
1.6 and 1.7 above. The values of variable x in equation stock prices
1.5, 1.6 and 1.7 were determined using 3-monthly moving
average period as shown in equation 1.8. Figure 1.2 shows
the predicted stock market prices for the months of July,
August, September, October, November and December,
year 2008 for the three banks. The PE shown in figure 1.2
columns 5, 7 and 9 are the equivalent values of variable x
obtained using the method of moving average as discussed
earlier. These values were further substituted into the
linear functions in equations 1.5, 1.6 and 1.7 to predict the
new market prices for each of the three banks.
Figure 1.3 (line graph) and 1.4 (bar chart) depict
the trends of the predicted stock market prices for the three
banks over a period of six months.

Figure 1.4: A bar chart showing trends of the predicted stock

prices

4. CONCLUSION
We presented regression analysis as a data
mining technique and developed tool for exploiting
especially time series data in financial institution. A
prediction system has been built that uses data mining
technique to produce periodically forecasts about stock
market prices. Our technique complement proven numeric
Figure 1.2: Predicted Stock Market Prices forecasting method using regression analysis with
technology taking as input the financial information

156
 Volume 1 No. 4, JULY 2011 ISSN 2222-9833
ARPN Journal of Systems and Software

©2010-11 AJSS Journal. All rights reserved

http://www.scientific-journals.org

obtained from the daily activity summary (equities) York: John Wiley and Sons, Inc. ISBN: 0-471-
published by Nigerian Stock Exchange. In this paper, we 66657-2
were able to use regression analysis as a data mining
technique to describe the trends of stock market prices and [7] Portia, A.C.(n.d). Data Mining and Neural
predict the future stock market prices of three banks as a Networks from a Commercial Perspective.
case study from banking sector of Nigerian economy. Australia: University of Technology Sydney,
Australia.
REFERENCES
[8] Two Crows Corporation (2005). Introduction to
[1] Arora, P.N. and Arora, S. (2009). Statistics for Data Mining and Knowledge Discovery, (3rd ed.).
Management. New Delhi: S. Chand & Company Potomac, USA: Two Crows Corporation. ISBN:
Limited. ISBN: 81-219-2285-2. 1-892095-02-5

[2] Fayyad, U., Piatetsky-Shapiro, G. and Smyth, P. [9] Weiss, G.M., and Davison, B.D. (2010). Data
(1996). From Data Mining to Knowledge in Mining. Handbook of Technology Management,
Databases. Menlo Park, Calif: American H. Bidgoli (Ed.), John Wiley and Sons.
Association for Artificial Intelligence.
[10] Wuthrich, B., Cho, V., Leung, S., Permunetilleke,
[3] Ravindranath, B. (2003). Decision Support D., Sankaran, K., Zhang, J., and Lam, W. (1998).
Systems and Data Warehouses. New Delhi: New Daily Stock Market Forecast from Texual Web
Age International Limited. ISBN: 81-224-1454-0. Data.

[4] Kannan, K.S., Sekar, P.S., Sathik, M.M. and

Arumugan, P. (2010). Financial Stock Market
Forecast using Data Mining Techniques. Hong Abdulsalam Sulaiman Olaniyi is a
Kong: Proceedings of International graduate of University of Ilorin, Ilorin,
MultiConference of Engineers and Computer where he obtained Bachelor of Science
Scientist 2010 Vol. I. IMECS 2010. ISBN: 978- Degree in Computer Science and Master of
988-17012-8-2. Science Degree in Computer Science
respectively. His areas of specialization include Software
[5] Han, J. and Kamber, M. (2006). Data Mining: Engineering and Data Mining. He is a member of Nigeria
Concepts and Techniques, (2nd ed.). Morgan Computer Society (MNCS). His is presently an Assistant
Kaufmaun. Lecturer in the Department of Physical Sciences, Al-
Hikmah University, Ilorin, Nigeria. The co-authors,
[6] Larose, D.T. (2005). Discovering Knowledge in Adewole, K. S. and Jimoh, R. G. are both Lecturers from
Data: An Introduction to Data Mining. New Kwara State University, Malete, Nigeria and University of
Ilorin, Ilorin, Nigeria respectively.

157

View publication stats