Volume 6, No.

6, July-August 2015
International Journal of Advanced Research in Computer Science
RESEARCH PAPER
Available Online at www.ijarcs.info

Time Series Data Analysis for Stock Market Prediction using Data Mining Techniques with R
Mahantesh C. Angadi Amogh P. Kulkarni
M.Tech Student, Dept. of ISE, Assistant Professor, Dept. of ISE,
Acharya Institute of Technology, Sai Vidya Institute of Technology,
Bengaluru, Karnataka, India. Bengaluru, Karnataka, India.

Abstract: Nowadays, the stock market is attracting more and more people's notice with its high challenging risks and high return over. A stock
exchange market depicts savings and investments that are advantageous to increase the effectiveness of the national economy. The future stock
returns have some predictive relationships with the publicly available information of present and historical stock market indices. ARIMA is a
statistical model which is known to be efficient for time series forecasting especially for short-term prediction. In this paper, we propose a model
for forecasting the stock market trends based on the technical analysis using historical stock market data and ARIMA model. This model will
automate the process of direction of future stock price indices and provides assistance for financial specialists to choose the better timing for
purchasing and/or selling of stocks. The results are shown in terms of visualizations using R programming language. The obtained results reveal
that the ARIMA model has a strong potential for short-term prediction of stock market trends.

Keywords: Time Series Data, Stock Market, Prediction, Analysis, Data Mining, ARIMA, R.

utilized among analysts and data excavators for statistical

I. INTRODUCTION
programming and data analysis. R language was created
Today we live and breathe data. Forecasting the stock by Ross Ihaka and Robert Gentleman at the University of
exchange data is an important financial subject which involves Auckland, New Zealand and is presently maintained by the R
an assumption that the fundamental information publicly development core team. Amid the most recent decade, the
available in the past has some predictive relationships to the energy originating from both the scholarly world and industry
future stock returns. Stock market forecasting contains has lifted the R programming dialect to turn into the most
uncovering the market trends, planning investment tactics, essential tool for computational insights, perception and data
identifying the best time to purchase the stocks and which science. Around the world, a large number of analysts and
stocks to purchase. A stock exchange or equity business sector information researchers use R language to take care of their
is a non-direct, non-parametric framework that is difficult to most difficult issues in the fields going from computational
model with any sensible exactness. It is the mix of speculators science to quantitative promoting.
who need to purchase or offer or hold a share at a specific time. R-Studio is a free and powerful integrated development
Prediction will continue to be an exciting locale of research, environment for R language. R-Studio allows the user to run R
making scientists in the analytics field always desiring to scripts in a more user-friendly environment. R-Studio is
enhance the existing forecasting models. The motivation is that available in two editions: R-Studio Desktop, where the program
companies and individuals are empowered to make investment is run locally as a regular desktop application; and R-Studio
decisions to develop viable system about their future Server, which allows accessing R-Studio using a web browser
endeavors. while it is running on a remote Linux server. R-Studio is
Stock trend forecasting is considered as one of the most written in the C++ programming language and uses the Qt
difficult tasks to achieve in money related gauging because of framework for its graphical user interface. R-Studio has a
the difficulty in the multifaceted world of stock market. Many general console where we can type commands and see output,
of the investors in the stock market are finding a technique that it has an editor that supports direct code execution and has
could guarantee easy profiting by forecasting the stock trends features of highlighting the syntax, debugging and managing
and minimize the risk of investing. This motivates the the workspace. R-Studio also has a workspace tab which
researchers in the domain field to delve and develop new includes all the active objects, a history tab which shows a list
forecasting models. Time series data analysis techniques use of commands used so far, and the files tab that shows all the
verifiable information as the premise for evaluating future files and directories in the default workspace.
results. Time series data can be defined as numerical data The remainder of the paper is organized as follows. Chapter
collected in a particular sequence over a period of time at 2 discusses about the literature review of previously published
regular intervals. The time series data can include the values research papers related to forecasting the stock trends. Chapter
collected at the end of every week, month, quarter, or year etc. 3 discusses about the system analysis. Chapter 4 discusses the
The intention is to find if there is any link between the data implementation methodology which is used to predict the stock
collected so far and in what way does the data changes. In order trends. Chapter 5 includes some of the screen shots of predicted
to reduce the risk of investment, exchange of securities results. We conclude the paper in chapter 6 by providing the
between the seller and buyer are facilitated by the stock ideas for future enhancements in chapter 7. Finally references
exchanges. A stock exchange is an organization or a place have been included.
where the stock traders or investors can deal with stocks. Some
of the examples for stock market organizations include II. LITERATURE REVIEW
NASDAQ, NYSE, BSE, NSE etc.
R is a programming language and environment for To attain the objective of forecasting the stock returns,
statistical processing and graphics. The R dialect is generally fundamental analysis and technical analysis techniques are used

© 2015-19, IJARCS All Rights Reserved 104

 Mahantesh C. Angadi et al, International Journal of Advanced Research in Computer Science, 6 (6), July-August, 2015, 104-108

by number of researchers. These stock trading rules are NSE. Artificial Neural Networks (ANNs) model is very
developed based on the information related to popular due to its ability to learn patterns from data and infer
macroeconomics, ancient data of stock trading. Fundamental solution from unknown data. Hybrid approaches also engaged
analysis techniques assume that the price of a stock depends on to improve stock price predictive models by exploiting the
its intrinsic values and expected return on investment. Majority unique strength of each of them. The results obtained from real-
of the researchers believe that fundamental analysis techniques life data demonstrated the potential strength of ARIMA models
are good methods only on a long-term source. Yet, for short- to provide investors short-term prediction that could aid
term and medium-term speculations technical analysis investment decision making process.
techniques are preferred as fundamental analysis techniques are Stock exchange markets facilitate savings and investment
not appropriate. that are beneficial to increase the effectiveness of national
The most efficient way to forecast the future is to economy. The author [Li Zhe] [7] have used the method of
understand the present scenarios. The author [Banerjee D] [1] technical analysis in which trading rules were established based
tried to develop an appropriate model that helps to forecast the on the ancient data of stock trading price and volume.
unseen values of the Indian stock market, based on the Technical analysis uses various methods that aim to predict
information collected on the monthly closing stock indices. future stock price movements based on the assumption that
Based on the ARIMA model they predict the future stock history repeats itself and future market directions can be
indices which have the strong performance of the Indian determined by examining historical stock prices.
economy. It is very important to understand the present status
III. SYSTEM ANALYSIS
of the market because for many economists, investors and
researchers the Indian stock market is the center of interest. It A. Problem Statement
has been predicted that the performance of the Indian stock
A stock exchange market depicts savings and investments
market presents a suitable time series ARIMA(1,0,1) model
that are advantageous to increase the effectiveness of national
which helps to create the appropriate values of the future
economic. The future stock returns have some predictive
indices.
relationships with the publicly available information of present
The authors [Li Bing] [Chan] [2] have extracted the
and historical stock market indices. The investors decide the
ambiguous text through MLP techniques to get the real stock
better time to sell/buy/hold a share in stock market based on the
price movements and public sentiments. It has been said that
former relationship. Every investor is interested in predicting
public emotions may be co-related that has shown through
the future stock prices, whether the investor may be a long-term
Twitter. The authors have used data mining algorithms to mine
investor or a day-trader. This possesses a major challenge to
Twitter data in order to forecast the stock trends using
design and develop an effective and efficient predictive model
sentimental analysis which comes under fundamental analysis.
that assists the investors to take appropriate decisions.
The estimation of the latest social media analysis works on up-
to date public views mining. Through this methodology for B. Existing Systems
social media data mining, it has fulfilled the recognized One of the significant financial subject that has engrossed
research problems with adequate experimental results. The the researcher’s attention for many years is forecasting the
algorithm has defined the complete relationships surrounded in stock returns. Investors in the stock market have been
social media as a graph with several layers. The top layered attempting to discover an answer to estimate the stock trends in
attributes and intermediate layered attributes have direct order to decide the better timing to buy or sell or hold a share.
relations; and bottom layered attributes and intermediate Forecasting the stock trends have been done both on qualitative
layered attribute have in-direct relations. analysis and quantitative analysis. There are many statistical
To forecast stock price trend the authors [Tao Xing] [Yuan models available for forecasting stock trends and choosing an
Sun] [3] have introduced a method based on Hidden Markov appropriate model for a particular forecasting application
Model. Hidden Markov Model first proposed by Baum and depends on the format of the data.
Egon, which is a kind of Markov Chain and is used for the
pattern recognition technique. This paper finds the hidden C. Proposed Study
relationship existing between the Hidden Markov Model and In this work we propose a prediction model for the time
stock prices. The experimental results show that, this method series stock market data. This model will automate the process
can get attractive accurate result, particularly efficient in short of change of stock price indices based on technical analysis and
period prediction. provides assistance for financial specialists to choose the better
It is tedious task for the stock market financial specialists to timing for purchasing and selling stocks. Data mining
guesstimate the pattern of the stock exchange costs as techniques are used to develop the prediction model and R
effectively as could be allowed to settle on the best exchanging programming language is used for visualization of results.
choices. The authors [Vishwanath R Ha] [Leena Sa] [4] have
proposed a system called APST, which performs the pre- IV. IMPLEMENTATION
processing of verifiable stock time arrangement information to Data mining can be interpreted as a knowledge discovery
produce the grouping of approximated values by utilizing
process. Data mining techniques are devised to address the
multi-scale segment mean methodology. To locate the closest
problems by providing a reliable model with data mining
neighbor objects they utilize the Euclidian separation way to
recognize the comparative arrangement of articles. The features. To construct a model that investigates the stock
experimental results of this system show that the executed patterns by utilizing the past stock exchange trends; we use the
framework has shown 74% of the exactness. auto-regressive integrated moving average (ARIMA) model.
The complete architecture of the system is shown below.

The authors [Ayodele A] [Adebiyi] [5] have used the

ARIMA model to develop an extensive process of building
stock price predictive model by obtaining data from NYSE and

© 2015-19, IJARCS All Rights Reserved 105

 Mahantesh C. Angadi et al, International Journal of Advanced Research in Computer Science, 6 (6), July-August, 2015, 104-108

C. Data Pre-processing: Data Wrangling

The data pre-processing stage involves all the activities to
prepare the final dataset from the preparatory raw information.
The data preparation tasks can be performed several times as
there is no specific order. These tasks include the selection of a
record, table, attribute and cleaning of data for modeling tools.
In our methodology, the input data will be converted into a
combined value vector list or differentiated value vector list.
So for this purpose we use c{base} that refers to the combined
values vector or list.
• Data Frames
A list of variables with same number of rows and distinct
row names of a class is defined as a data frame. The row
names decide the number of rows, if no variables are involved.
The essential and default data structure in R language is
‘data.frames{base}’. Data frames are the collection of
variables which are tightly coupled with many properties of
lists and matrices. The data frame alters each and every
argument data.frame by calling as data.frame(optional= true).
The behavior of the arguments can be changed by writing
Fig 4.1 System architecture methods according to their classes and also R consists of many
System architecture is a model that defines the behavior of such methods.
a system in the conceptual model. The huge systems are D. Data Processing: Data Training
decomposed into subordinate systems to provide similar set of To process the data we use ARIMA(p,d,q) model. In
services. The beginning layout strategy of perceiving these technical analysis investors use the auto regressive and
sub-systems and building up a structure for sub-systems moving average models to forecast the stock trends. Major
control and cooperation is called architecture design. As steps involved here are identification, parameter estimation
shown above, Fig. 4.1 includes seven major steps to and forecasting. These steps are repeated until an appropriate
implement the system and each step is explained below. model is identified for prediction. R provides auto.arima()
A. Understanding the Objective method to forecast the time series data according to
The first step in developing a project is to understand the ARIMA(p,d,q).
objective which involves an understanding of the intent and • Auto Regression (AR)
essentials of a system. This comprehension is used as a
problem description and a preparatory system to accomplish Auto regression technique estimates the future values
the expectations. The objective of our project is neither to build based on the previous values. The first order of auto regression
a system that makes billions nor to waste billions too. But the process is represented as AR(1), meaning that the immediately
objective is to develop a system that finds the direction of preceding values will be based on current value. The current
change of stock price indices based on the co-relations between value based on the previous two values is an AR(2) process.
stock prices and help the investors in the stock market in taking The word auto regression says that it is a regression of the
a decision whether to buy/sell/hold a stock by providing the variable against itself.
results in-terms of visualizations. • Moving Average (MA)
B. Data Collection A moving average is a technique to find the overall idea
Once the understanding of the objective is over, the next within a data set. It finds the future trends based on the
step is to collect the data. Data collection involves the previous values. The two commonly used moving average
understanding of initial observations of the data to identify the techniques are exponential moving average (EMA) and the
useful subsets from hypotheses of the hidden information. simple moving average (SMA).
Here we use R script to collect the data from Google finance • Order of ARIMA
using getSymbols() method of QuantMod package.
The order of an ARIMA model is generally represented as
• QuantMod
ARIMA(p,d,q), where-
It refers to the Quantitative financial Modeling which is
used in financial time series data analysis. QuantMod is the p = order of the autoregressive part
rapid prototyping environment where the quant traders can d = degree of first differencing involved
quickly create and discovery the trading models. To extract q = order of the moving average part
and load the data from multiple sources we use a method Here if d=0, then the model becomes ARMA which is linear
called getSymbols(). As a source for obtaining the stock stationary model.
market data, most of the stock investors use Google finance or The same stationary and in-variability conditions that are
Yahoo finance. We used Google finance to load the data into used for autoregressive and moving average models apply to
R by specifying the correct ticker symbol. The OHLC data is this ARIMA(p,d,q) model. Selecting the appropriate values for
not directly downloaded from the Google finance p, d and q can be challenging. The auto.arima( ) function in R
(finance.google.com), instead a call to getSymbols(Symbols, will do it automatically.
src='google') in-turn call this method.

© 2015-19, IJARCS All Rights Reserved 106

 Mahantesh C. Angadi et al, International Journal of Advanced Research in Computer Science, 6 (6), July-August, 2015, 104-108

• Model Estimation for ARIMA F. Plot Visualizations

Model estimation for ARIMA can be achieved based on Data visualization is a graphical representation of the
the pre-processed historical data. numerical data. In our methodology, after forecasting the stock
market trends we visualize the results for short-term
investment assistance in-terms of line charts, candlesticks
charts, bar charts, and histograms. Here x-axis shows the time
period in-terms of year/months/days and y-axis shows the
stock price values.
G. View and Analyze Results
Once after plotting the results in-terms of visualizations we
can find out the correlations to get the short-term predictions.
In the next section we provide some of the screen shots by
which the investor can analyze and predict the future stock
trends of a particular company at a specific time period. So the
investors in the stock market can use this as assistance to
sell/buy/hold a share.
V. SCREEN SHOTS
Fig. 4.2 Model Estimation for ARIMA(p,d,q)
In ARIMA model, the identification is to be accomplished
using auto co-relation function and partial auto co-relation
function in order to identify p, d and q standards. For any
realistic time sequence generally p, d and q values vary
between 0 and 2, but model estimation is executed for all
probable combinations of p, d and q values. The pictorial
representation of these steps is shown in Fig. 4.2.
• ARIMA() Function in R
The auto.arima() is a very useful function in R, but
anything automated can be a little dangerous. This function
examines over conceivable models within the edict limitations
provided and returns the best ARIMA model. The value
of d also has an effect on the prediction intervals i.e., the more
complex the value of d, the more rapidly forecasting intervals
surge in size. For d=0, the long-term prediction average
deviance will go to the typical deviance of the historic data. It Table. 5.1 Intra-day INFY data from Google finance - 2007 to 2015
is usually not possible to tell merely from a time plot, what
values of p and q are suitable for the specific kind of data.
Sometimes it is conceivable to use the ACF plot and closely
related PACF plot to govern the appropriate values.
• Special Cases of ARIMA Model

Fig. 5.1 Chart series graph for INFY data - 2007 to 2015

Table 4.1 Special Cases of ARIMA Model

E. Forecasting Results
The process of making predictions of the future by
relying upon the past and present data is known as forecasting.
Various prediction techniques are used by the stock analysts to
evaluate the future stock trends value. Prediction also offers a
significant standard for organizations that have a long-term
perception of actions. We use ‘forecast’ package for predicting
the future stock trends based on the analysis of past trends.
This ‘forecast’ package provides a number of forecasting
functions for displaying the time series predictions along with
exponential smoothing and space models. Fig. 5.2 Line chart graph for INFY data – 2014 to 2015

© 2015-19, IJARCS All Rights Reserved 107

 Mahantesh C. Angadi et al, International Journal of Advanced Research in Computer Science, 6 (6), July-August, 2015, 104-108

REFERENCES
[1] Banerjee, D., "Forecasting of Indian stock market using time-
series ARIMA model", 2nd IEEE International Conference on
Business and Information Management (ICBIM), January 2014,
pp. 131-135.
[2] Li Bing, Chan, K. C. C., C. Ou, "Public sentiment analysis in
Twitter data for prediction of a company's stock price
movements", 11th IEEE International Conference on e-Business
Engineering (ICEBE), November 2014, pp. 232-239.
[3] Tao Xing, Yuan Sun, Qian Wang, Guo Yu, “The analysis and
prediction of stock prices”, IEEE International Conference on
Granular Computing (GrC), December 2013, pp. 368–373.
Fig. 5.3 Re-chart graph for INFY Data - First 16 Weeks of 2007 [4] Vishwanath R. Ha, Leena Sa, Srikantaiah K. Ca, K. Shreekrishna
Kumar b., P. Deepa Shenoya, Venugopal K. Ra, S. S. Iyengarc,
L. M. Patnaik, "Forecasting stock time-series using data
approximation and pattern sequence similarity", International
Journal of Information Processing (IJIP), September 2013, pp.
90-100.
[5] Ayodele A. Adebiyi, Aderemi O. Adewumi, Charles K. Ayo,
"Stock price prediction using the ARIMA model", 16th IEEE
International Conference on Computer Modelling and
Simulation (UKSim), March 2014, pp. 106 -112.
[6] Qasem A. Al-radaideh, Adel Abu Asaf, Eman Alnagi,
“Predicting stock prices using data mining techniques”, The
International Arab Conference on Information Technology 2013.
[7] Li Zhe; “Research on China's stock exchange markets: problems
Fig. 5.4 Forecasted INFY Trends – Line chart graph 2014 to 2015 and improvements”, International Conference on Education and
Management Technology, 2010. pp 465-469.
[8] Hazem M. El-Bakry, and Wael A. Awad, “Fast forecasting of
stock market prices by using new high speed time delay neural
networks”, International Journal of Computer and Information
Engineering, February 2010, pp. 138-144.
[9] Han, J., Kamber, M., Jian P., “Data mining concepts and
techniques”. San Francisco, CA: Morgan Kaufmann Publishers,
2011.
[10] Wang, Y.F., “Mining stock price using fuzzy rough set system”,
Expert Systems with Applications, 2006, pp. 13-23.
[11] Enke, D., Thawornwong, S., “The use of data mining and neural
networks for forecasting stock market returns”, Expert Systems
with Applications, 2005, pp. 927-940.
Fig. 5.5 Forecasted INFY Trends – Histogram graph 2014 to 2015 [12] A. J. Conejo, M. A. Plazas, R. Espnola and B. Molina. “Day-
ahead electricity price forecasting using the wavelet transform
VI. CONCLUSION and ARIMA models”, IEEE Transactions on Power Systems,
2005, pp. 1035–1042.
In this paper we made an attempt to develop a prediction
model for forecasting the stock market trends based on the
technical analysis using historical time series stock market data SHORT BIO DATA OF AUTHORS
and data mining techniques. The experimental results obtained Mahantesh C. Angadi: He obtained his B.E (Information
demonstrated the potential of ARIMA model to predict the
Science and Engineering) from Visvesvaraya Technological
stock price indices on short-term basis. This could guide the
investors in the stock market to make profitable investment University, Belgaum, Karnataka, India. Currently he is
decisions whether to buy/sell/hold a share. With the results pursuing final year M.Tech (Computer Networking and
obtained ARIMA model can compete reasonably well with Engineering) in Acharya Institute of Technology, Bengaluru,
emerging forecasting techniques in short-term prediction. Karnataka, India. His areas of interest include BigData
analytics using Hadoop and R, Distributed computing, Cloud
computing and Networking.
VII. FUTURE ENHANCEMENTS
Amogh P. Kulkarni: He obtained his M.Tech (Software
The implementation of this paper can be extended by Engineering) from R.V College of Engineering, Visvesvaraya
integrating the technical analysis and fundamental analysis Technological University, Belgaum, Karnataka, India.
techniques. Through the evaluation of social media analysis Currently he is working as an assistant professor in Sai Vidya
particularly on public opinions using fundamental analysis Institute of Technology, Bengaluru, Karnataka, India. His
techniques can be incorporated in order to obtain better results. areas of interest include BigData analytics, Distributed
In this way we can provide the improved results for investors in computing, Cloud computing and Networking.
the stock market to choose the better timing for profitable
investment decisions.

© 2015-19, IJARCS All Rights Reserved 108

Copyright of International Journal of Advanced Research in Computer Science is the property
of International Journal of Advanced Research in Computer Science and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.