1 2 3
Sridevi. U.K , Shanthi Palaniappan , Nagaveni Palanisamy
1
Department of Computer Applications,
Sri Krishna College of Engineering and Technology, Coimbatore, INDIA
2
Department of Computer Applications,
Sri Krishna College of Engineering and Technology, Coimbatore, INDIA
3
Department of Mathematics
Coimbatore Institute of Technologies, Coimbatore, INDIA
ABSTRACT
Based on the past data the company runs several promotional markdown events
throughout the year. The data analytics helps in business to build better
products and deliver better services. Predictive analytics will make a
personalized recommendation of products. Data analyses have been growing
importance on the stock market in the recent years. The quarterly by-store profit
information is given and the prediction is done for the following year’s profit
based on the different environment variables that affect the profit of the store.
The seasonal variations can be identified by additive and multiplicative models.
The model is developed for profit prediction. The accuracy of the model is
compared with the results of the forecast error calculation. The results show that
the accuracy of the model is better than that of other model.
1 INTRODUCTION
1931
International Journal of Pure and Applied Mathematics Special Issue
An online retailer announces the “Big Sale Day” for a particular week. The
trend in purchase many vary and the discount for the item may be fixed based on
trend analysis, the purchased may be increased during the discount week and in
the same way after the week, the business may be less. The prediction of market
analysis needs to be done the company to improve their sales.
2. PROBLEM OBJECTIVE
3. RELATED WORK
Data analyses have been growing importance on the stock market in the
recent years. In order to get the profit of the investing, many investors need to
know how to analyze the important data from the stock market. The interest for
data mining techniques has increased tremendously during the past decades,
and numerous classification techniques have been applied in a wide range of
business applications (Thomas, 2013). Predictive performance evaluation is a
fundamental issue in design, development, and deployment of classification
systems (Huang, 2015). Predictive performance evaluation is considered as a
multidimensional problem. The single scalar summaries such as error rate can
be used to evaluate all the aspects that a complete and reliable evaluation must
be considered (Ronaldo, 2011). Thomas (2013) developed a predictive model
which maximizes the profit and classification accuracy. Data mining techniques
have been applied to stock (market) prediction. The pre-processing step of data
mining filters out unrepresentative variables from a given dataset for effective
prediction (Chih-Fong Tsai,2010). Zhang et al., used the predictive model to
forecast the stock market through twitter.
1932
International Journal of Pure and Applied Mathematics Special Issue
The method can be used to curve fitting , designing a system around past
data rather than identifying repeatable behavior, represents a potential risk for
company the company should seek parameter that correlate to changes in the
price of a given security. The technical indicator is a series of data points that
are derived by applying formula to the prices data of security includes any
combination of the open, high, low or close over a period of time. The technical
indicator used are SMA (Simple Moving Average), a simple or arithmetic,
moving average that is calculated by adding the closing price of the security for a
number of time periods and then dividing this total by number of time periods.
MACD (Moving Average Convergence Divergence) MACD indicator is one of the
simplest and the most effective available momentum indicator. The MACD turns
two trends following indicator, moving averages into momentum oscillator by
subtracting the longer moving average from the shorter moving average. Chih-
Fong Tsai(2010) used the multiple feature selection in the prediction. The
moving average and weighted moving average method can be used in forecasting
(Hatchett, 2009). Lijun et al., used genetic algorithm is applied to evaluate the
moving average trading rules on future market prediction.
4. METHODOLOGY
1933
International Journal of Pure and Applied Mathematics Special Issue
(b) Use the trend line to forecast future trend line values.
(c) Adjust these values by the average seasonal variation applicable to the future
period to determine the forecast for that period.
Moving Average
The simple moving average is mainly used by online traders. The trend
directions can be identified and it produces signals to buy and sell.
Consider the prices for the last 5 days were 25, 28, 26, 24, 25.
Assign weight for the recent day as the weight of 3, the second recent day may
be 2, and the last day of the period receives a weight of 1.
The weights are assigned in decreasing order and the sum of the weight is 1.
1934
International Journal of Pure and Applied Mathematics Special Issue
The store profit information is given for last 4 years of the stores and the
prediction needs to be done for the following year’s profit for shops based on
environmental variables of the stores. The quarterly profits needs to be predicted
for the stores. The moving averages could cover the profits of a shop over periods
of four quarters. The forecast error is used to calculate the accuracy of the model
and the sample forecast accuracy is given in the table 1. The various parameters
that influence the profit of the shop is listed in the table 2.
Weighing the features minimize the error rate improves the prediction
accuracy by at least 10% in the all the test cases. The trends analysis is
identifies both the profit and sales. This representation is useful for complex
nonlinear models with multiple inputs.
value of
forecast
Series
Week
value
Time
error
error
% of
1 17 - -
2 21 17 4 19.05
3 19 21 2 10.53
4 23 19 4 17.39
5 18 23 5 27.78
6 16 18 2 12.5
1935
International Journal of Pure and Applied Mathematics Special Issue
7 20 16 4 20
8 18 20 2 11.11
9 22 18 4 18.18
10 20 22 2 10
11 15 20 5 33.33
12 22 15 7 31.82
1936
International Journal of Pure and Applied Mathematics Special Issue
13 654 2181.92
14 654 17504.53
15 654 19403.03
16 654 -2877.26
The store’s profit is listed below in Table 3 and the moving average for the
interval 3 is in the Table 4.
1937
International Journal of Pure and Applied Mathematics Special Issue
The figure 1 shows the initial visual analysis of the series “Sales” from “data”. The
graph is generated by using RStudio forecasting library. The model is build with 80% of
dataset for training phase and the predicted with 20% of dataset for testing phase. The
prediction accuracy is based on the ratio of the correct predicted values to the testing
data observation.
5. Conclusion
The predictive model has implications for business in that sellers can utilize the
proposed system to effectively predict the profit. By predicting the profit of the
company can optimize the growth of the company. Based on the prediction the
manager can develop plans for the products and infrastructure development.
6.Reference
1938
International Journal of Pure and Applied Mathematics Special Issue
[4] Huang, W.,Qing Zhang., Wei Xu., Xun Liang, “A Novel Trigger Model for
Sales Prediction with Data Mining Techniques”. Data Science Journal, 14,
2015
[5] Lijun Wang., Haizhong An., Xiaohua Xia., Xiaojia Liu., Xiaoqi Sun., and
Xuan Huang, “ Generating Moving Average Trading Rules on the Oil
Futures Market with Genetic Algorithms”, Mathematical Problems in
Engineering, 2014
1939
1940