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

net/publication/383864006

Retail Sales Prediction Using Machine Learning Algorithms Retail Sales

Prediction Using Machine Learning Algorithms

Article · September 2024

CITATIONS READS

2 531

5 authors, including:

Rao Srinivasa Bandaru Kamepalli Sujatha

Vignan University VFSTR
89 PUBLICATIONS 370 CITATIONS 153 PUBLICATIONS 844 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Rao Srinivasa Bandaru on 09 September 2024.

The user has requested enhancement of the downloaded file.

 Retail Sales Prediction Using Machine Learning Algorithms

Turkish Online Journal of Qualitative Inquiry (TOJQI)

Volume 12, Issue 1, Janurary 2021: 315-322

Research Article

Retail Sales Prediction Using Machine Learning Algorithms

Dr. Bandaru Srinivasa Rao1
Dr. Kamepalli Sujatha2
Dr. Nannpaneni Chandra Sekhara Rao3
Mr. T. Nagendra Kumar4
1, 2, 4
Vignan Foundation for Science, Technology and Research (VFSTR),
Deemed to be University, Vadlamudi, Guntur District, Andhra Pradesh, India.
3
VR Siddhartha Engineering College, Vijayawada, Andhra Pradesh, India.

Abstract

Data and Data Science, both are playing a pivotal role in creating business intelligence. Machine
learning became buzzword technology in data processing and analytics. Retailors in an unorganized
sector take decisions intuitively whereas organized retail concerns use business intelligence for their
business decisions. Retail sales prediction is common in the organized retail sector, which is highly
useful for in time and strategic competitive decisions. In general, sales predictions are made based on
past data using statistical and mathematical techniques, etc. The present paper is an attempt to predict
retail sales using machine learning algorithms and to present the accuracy of results to the retailors,
managers, and policymaker in the retail industry. In this paper three machine learning algorithms k-
Nearest Neighbour regression model, Multinomial regression model, Ada Boost regression model
were implemented on the training dataset. Multinomial and Decision Tree with Ada Boost regression
models work with 100% accuracy. Decision Tree with Ada Boost regression model was
implemented on the test set to predict the outlet sales. The regression model predicted the sales of a
particular item indicated with the item identifier in a particular outlet indicated with the outlet
identifier on the test set. The implications for retailers, managers, and policymakers in the retailing
sector are provided based on the results.

Keywords: Retail, Sales, Prediction, Machine Learning Algorithms

1. Introduction

Business intelligence became a buzzword in the 21st century. Intuitional decision making is replaced
by business intelligence in the Industrial evolution 4.0. Most of the economic decisions are
happening to take into consideration data, data processing, results, and analytics. Information
Technology plays a pivotal role in reducing data processing time and assurance of precession,
reliability, and accuracy of results, which is highly useful for valuable predictions and business
decisions. The retail sector is a combination of unorganized and organized concerns. Most of the
unorganized retail concerns take decisions intuitively or based on the experience of proprietors
whereas organized retail concerns prefer to make decisions based on data/information process
results. Organized Food and grocery sector is playing a dynamic role in penetrating/occupying into

315
 Dr. Bandaru Srinivasa Rao, Dr. Kamepalli Sujatha, Dr. Nannpaneni Chandra Sekhara Rao,
Mr. T. Nagendra Kumar
more and more unorganized food and grocery market. Organized food and grocery sector concerns
are using contemporary technology for processing data to get better results in less time and applying
data science for analytics for transforming the data into business intelligence for using a competitive
advantage. In this scenario, academicians, researchers and practitioners contributions for retailers
knowledge are: sales prediction in super markets (Thiesing, Middelberg, & Vornberger, 1995); food
sales prediction (Meulstee & Pechenizkiy, 2008); simulation based sales forecasting in retail sector
(Lv, Bai, Yin, & Dong, 2008), (Schwenke, Ziegenbalg, & Dresden, 2012) forecasting retail sales
using neural network in (Y. F. Gao, Liang, Liu, Zhan, & Ou, 2009); using differential evolution (R.
Majhi, Panda, Majhi, Panigrahi, & Mishra, 2009), neural network and VBA (Y. Gao, Liang, Tang,
Ou, & Zhan, 2010), algorithm (Y. Gao, Liang, Zhan, Ren, & Ou, 2011), Linear Regression
(Gopalakrishnan, Choudhary, & Prasad, 2018), Data Mining Techniques (İşlek, 2015),(Singh,
Ghutla, Jnr, Mohammed, & Rashid, 2017), (Gaku & Takakuwa, 2015), two-Level Statistical Models
(Punam, Pamula, & Jain, 2018), Grid Search Optimization ( GSO) (Behera & Nain, 2019),
Supervised Algorithms (Ohrimuk & Razmochaeva, 2020), machine learning techniques (Tsoumakas,
2018), (Arif, Sany, Nahin, Shahariar, & Rabby, 2019), (Krishna, Akhilesh, Aich, & Hegde, 2018),
(Wang & Liu, 2019), Time Series (Ping, 2018), Xgboost technique (Behera, 2019), Neural
Networks & GAS (Baba, Science, City, Prefecture, & Suto, 2000), PSO based adaptive ARMA
model (B. Majhi, 2009), S V R mathematical model and methods (Yang & Huiyov, 2007), Structure
Time Series (Wei, Geng, Ying, & Shuaipeng, 2014), Machine Learning Techniques (Cheriyan,
Ibrahim, & Treesa, 2018), (Wu, Patil, & Gunaseelan, 2018), Fuzzy Inference System (Liang, Li, &
Chen, 2019), strategies and technologies for marketers (B. Srinivasa Rao, 2018), recent applications
of machine learning (Sujatha Kamepalli, Bandaru Srinivasa Rao, 2018, 2019), praiseworthy. These
contributions disclose the role of mathematics, statistics, and econometrics and information
technology in providing business intelligence for competitive advantage.

Machine learning applications in data science are increasing day-by-day due to its capability of
quick, precise, and accurate processing of data using mathematics, statistical and econometric
applications, and providing business intelligence for business strategies development and execution.
The present study is an attempt to efficiently predict retail sales using machine learning algorithms.
The secondary objective of the study is to explore state-of-art tools, techniques, methods, and models
used in sales prediction organized retail outlets. The primary objective is to predict organized
retail sales using machine learning algorithms and comparing the results from implemented models.
The scope of the study is limited to Big Mart sales prediction using machine learning algorithms.
The study is expected to useful for retailers, decision-makers working for retail concerns, and
policymaker for development, regulation, and control.

The work is organized by the identification of state-of-art technology for predicting sales in
organized retail outs with precession. It follows methodology, data pre-processing, and building
prediction model, implementing the prediction model on the training data set, identifying the best
model for prediction, and implementing that model on a test set to predict outlet sales. Finally, it
concludes the findings and implications.

2.Methodology
In this research paper, we used machine learning regressor models to predict the big mart outlet
sales. Here we implemented three models on the training set. The selected best model implemented

316
 Retail Sales Prediction Using Machine Learning Algorithms

on a test set to predict the outlet sales. The following diagram shows the flow of work followed in
sales prediction.

Experimentation

The experiments were conducted by developing a simulation environment in python also using
WEKA. Three machine learning algorithms k-Nearest Neighbour regression model, Multinomial
regression model, Ada Boost regression model were implemented on the training dataset.

Data Feature
Cleaning Extraction

Training Set Test Set Prediction

Results
Big Mart Sales
Dataset
Build the Implement Best
Prediction Prediction
Model Model

Model 1 Model 2
Model n

Fig 1. Proposed Flow of Work

Source: Proposed Flow of Work Represented Graphically

k-Nearest Neighbour (k-NN) Regression Model: The k-NN algorithm is a non-parametric strategy
used for regression analysis. In this model the input consists of k nearest training examples in the
feature space. The output value can be obtained by the average of k nearest neighbour values. In k-
NN algorithm it uses three distance measures Euclidean Distance, Manhattan Distance and
Minkowski Distance.

The Euclidean Distance can be represented as

k
2
Distance = √∑(xj − yj )
j=1

The Manhattan Distance can be represented as

k

Distance = ∑|xj − yj |
j=1

The Minkowski Distance can be represented as

317
 Dr. Bandaru Srinivasa Rao, Dr. Kamepalli Sujatha, Dr. Nannpaneni Chandra Sekhara Rao,
Mr. T. Nagendra Kumar
1/q
k
q
Distance = (∑(|xj − yj |) )
j=1

Multinomial Regression Model: Multinomial Logistic Regression generalizes logistic regression to

multiclass issues. This model was utilized to foresee the probabilities of the various potential results
of a completely disseminated dependent variable, given a lot of independent factors. Multinomial
Logistic Regression models how multinomial reaction variable Y relies upon a lot of k informative
factors, X= (X1, X2, ... Xk).

Decision Tree Regression-Ada Boost Model: An AdaBoost regressor is a meta-estimator that starts
by fitting a regressor on the first dataset and afterward fits extra duplicates of the regressor on the
equivalent dataset yet where the loads of occasions are balanced by the error of the current forecast.
All things considered, resulting regressors focus more around troublesome cases.

Dataset Description: The dataset was collected from Analytics Vidhya web portal. It is of two parts
training set and testing set. The training set contains 8523 samples with 12 attributes and test set
contains 5681samples. The following table gives attribute description of the training set.

Table 1. Dataset Description

Source: https://www.analyticsvidhya.com/

The following diagram shows the attribute visualization of training set.

318
 Retail Sales Prediction Using Machine Learning Algorithms

Fig 2. Attribute Visualization of Training set

Source: Experimental Setup (WEKA3.8.3)

3. Results and Discussions

The three models k nearest neighbour Regression Model, Multinomial Regression Model and
Decision tree Regression- Ada Boost Model were implemented on training set. Three models work
with accuracy levels 0.85, 1.00, 1.00 respectively. The following table shows the accuracy of three
models on the training dataset.

Table 2. Comparison of Accuracy of Models

S. NO. Regression Model Accuracy

1 k nearest neighbour Regression Model 0.85
2 Multinomial Regression Model 1.0
3 Decision tree Regression- Ada Boost Model 1.0
Source: Experimental Results

The following diagram shows the graphical representation of comparison of accuracy of three
models.

Accuracy
1.05
1
0.95
0.9
0.85
0.8
0.75
k nearest Multinomial Decision tree
neighbour Regression Regression-
Regression Model Ada Boost
Model Model

Fig 3. Graphical Representation of Comparison of Accuracy of Models

Source: Experimental Results

319
 Dr. Bandaru Srinivasa Rao, Dr. Kamepalli Sujatha, Dr. Nannpaneni Chandra Sekhara Rao,
Mr. T. Nagendra Kumar
Multinomial and Decision Tree with Ada Boost regression models results in with an accuracy of
100%. So, Decision Tree with Ada Boost regression model was implemented on test set to predict
the outlet sales. The following table shows the sample output. Here we randomly considered some
outlets to visualize the output and we considered two digits after decimal point in the predicted
values.

Table 3. Sample Output from Decision Tree with Ada Boost regression model

S. No. Item Identifier Outlet Identifier Item_Outlet_Sales

1 FDW58 OUT049 1827.07
2 FDY38 OUT027 6373.43
3 FDC48 OUT027 2133.95
4 FDQ56 OUT045 1445.86
5 DRL59 OUT013 837.94
6 DRC12 OUT018 2973.97
7 FDG52 OUT046 788.68
8 FDX22 OUT010 499.05
9 FDE21 OUT035 1959.97
10 NCR06 OUT018 523.79
11 FDC26 OUT013 1837.78
12 NCS41 OUT035 3135.96
13 FDU34 OUT046 2088.55
14 FDM03 OUT013 1769.50
15 FDV44 OUT027 5146.45
16 FDT04 OUT013 611.97
17 FDW12 OUT035 2500.40
18 NCY42 OUT027 3495.73
19 FDA14 OUT013 2267.96
20 FDU58 OUT013 3087.11
21 FDB23 OUT046 3767.15
22 FDF47 OUT027 6236.36
23 NCA30 OUT045 3029.83
24 DRM48 OUT035 669.02
25 FDS08 OUT049 2729.96
26 FDC53 OUT019 263.36
Source: Experimental Results

From the above table it is clear that the regression model predicted the sales of a particular item
indicated with item identifier in a particular outlet indicated with outlet identifier. Hence this model
best predicts the outlet sales prediction. Using these results, Big Mart will try to understand the
properties of products and outlets which play a key role in increasing sales.

4.Conclusion
Machine learning applications are increasing day-by-day in business data processing and analytics

320
 Retail Sales Prediction Using Machine Learning Algorithms

area. In this work the results derived from the machine learning based algorithms are more precise,
accurate and confidently use for decision making than other models. Compared to three models the
last two models are giving hundred per cent accurate results. The results of this study can boost
confidence of retailers to implement machine learning in their business data processing and analysis.
It also useful for the managers associated with retail sector for developing suitable competitive
marketing strategies. Last but not least it helps the policy makers to make different estimations and
make suitable policies relating to retail sector.

References
1. Arif, A. I., Sany, S. I., Nahin, F. I., Shahariar, A. K. M., & Rabby, A. (2019). Comparison Study : Product Demand
Forecasting with Machine Learning for Shop.
2. B. Srinivasa Rao (2018), Butterfly Customers: Strategies and Technology for Marketers, International Journal of
Engineering & Technology, 7 (3.24) (2018) 512-516
3. Baba, N., Science, I., City, K., Prefecture, O., & Suto, H. (2000). for Constructing an Intelligent Sales Prediction.
565–570.
4. Behera, G. (2019). A Comparative Study of Big Mart Sales Prediction A Comparative Study of Big Mart Sales
Prediction. (October).
5. Behera, G., & Nain, N. (2019). Grid Search Optimization ( GSO ) Based Future Sales Prediction For Big Mart. 172–
178. https://doi.org/10.1109/SITIS.2019.00038
6. Cheriyan, S., Ibrahim, S., & Treesa, S. (2018). Intelligent Sales Prediction Using Machine Learning Techniques. 53–
58.
7. Dr. Sujatha Kamepalli and Dr. Srinivasa Rao Bandaru (2018) Implementation Framework of
8. Artificial Intelligence in Financial Services, International Journal of Research and Analytical Reviews, November
2018, Volume 5, Issue 4.
9. Gaku, R., & Takakuwa, S. (2015). Big data-driven service level analysis for a retail store. (2008), 791–799.
10. Gao, Y. F., Liang, Y. S., Liu, Y., Zhan, S. Bin, & Ou, Z. W. (2009). A neural-network-based forecasting algorithm
for retail industry. Proceedings of the 2009 International Conference on Machine Learning and Cybernetics, 2(July),
919–924. https://doi.org/10.1109/ICMLC.2009.5212392
11. Gao, Y., Liang, Y., Tang, F., Ou, Z., & Zhan, S. (2010). A demand forecasting system for retail industry based on
neural network and VBA. 2010 Chinese Control and Decision Conference, CCDC 2010, 3786–3789.
https://doi.org/10.1109/CCDC.2010.5498506
12. Gao, Y., Liang, Y., Zhan, S., Ren, X., & Ou, Z. (2011). Realization of a demand forecasting algorithm for retail
industry. Proceedings of the 2011 Chinese Control and Decision Conference, CCDC 2011, 4227–4230.
https://doi.org/10.1109/CCDC.2011.5968968
13. Gopalakrishnan, T., Choudhary, R., & Prasad, S. (2018). Prediction of Sales Value in online shopping using Linear
Regression. 5–10.
14. İşlek, İ. (2015). A Retail Demand Forecasting Model Based on Data Mining Techniques. 55–60.
15. Krishna, A., Akhilesh, V., Aich, A., & Hegde, C. (2018). Sales-forecasting of Retail Stores using Machine Learning
Techniques. 160–166.
16. Liang, Y., Li, J., & Chen, M. (2019). Online Shop Daily Sale Prediction Using Adaptive Network-Based Fuzzy
Inference System.
17. Lv, H. R., Bai, X. X., Yin, W. J., & Dong, J. (2008). Simulation based sales forecasting on retail small stores.
Proceedings - Winter Simulation Conference, (1968), 1711–1716. https://doi.org/10.1109/WSC.2008.4736257
18. Majhi, B. (2009). Efficient sales forecasting using PSO based adaptive ARMA model. 1333–1337.
19. Majhi, R., Panda, G., Majhi, B., Panigrahi, S. K., & Mishra, M. K. (2009). Forecasting of retail sales data using
differential evolution. 2009 World Congress on Nature and Biologically Inspired Computing, NABIC 2009 -
Proceedings, 1343–1348. https://doi.org/10.1109/NABIC.2009.5393740
20. Meulstee, P., & Pechenizkiy, M. (2008). Food sales prediction: “If only it knew what we know.” Proceedings - IEEE
International Conference on Data Mining Workshops, ICDM Workshops 2008, 134–143.
https://doi.org/10.1109/ICDMW.2008.128

321
 Dr. Bandaru Srinivasa Rao, Dr. Kamepalli Sujatha, Dr. Nannpaneni Chandra Sekhara Rao,
Mr. T. Nagendra Kumar
21. Ohrimuk, E. S., & Razmochaeva, N. V. (2020). Study of Supervised Algorithms for Solve the Forecasting Retail
Dynamics Problem. 441–445.
22. Ping, X. (2018). Particle Filter Based Time Series Prediction of Daily Sales of an Online Retailer.
23. Punam, K., Pamula, R., & Jain, P. K. (2018). A Two-Level Statistical Model for Big Mart Sales Prediction. 2018–
2021.
24. Schwenke, C., Ziegenbalg, J., & Dresden, D.-. (2012). Simulation based Forecast of Supermarket Sales Chair for
Technical Information Systems.
25. Singh, M., Ghutla, B., Jnr, R. L., Mohammed, A. F. S., & Rashid, M. A. (2017). Walmart ’ s Sales Data Analysis- A
Big Data Analytics Perspective. 114–119. https://doi.org/10.1109/APWConCSE.2017.00028
26. Sujatha Kamepalli, Bandaru Srinivasa Rao (2019), International Journal of Innovative Technology and Exploring
Engineering (IJITEE), ISSN: 2278-3075, Volume-8, Issue-6C2, April 2019
27. Thiesing, F. M., Middelberg, U., & Vornberger, O. (1995). Short term prediction of sales in supermarkets. IEEE
International Conference on Neural Networks - Conference Proceedings, 2, 1028–1031.
https://doi.org/10.1109/icnn.1995.487562
28. Tsoumakas, G. (2018). A survey of machine learning techniques for food sales prediction. Artificial Intelligence
Review. https://doi.org/10.1007/s10462-018-9637-z
29. Wang, J., & Liu, L. (2019). A Selection of Advanced Technologies for Demand Forecasting in the Retail Industry.
317–320.
30. Wei, D., Geng, P., Ying, L., & Shuaipeng, L. (2014). A Prediction Study on E-commerce Sales Based on Structure
Time Series Model and Web Search Data. i, 5346–5351.
31. Wu, C. M., Patil, P., & Gunaseelan, S. (2018). Comparison of Different Machine Learning Algorithms for Multiple
Regression on Black Friday Sales Data.
32. Yang, Y., & Huiyov, C. (2007). S V R mathematical model and methods for sale prediction. 18(4), 769–773.

322

View publication stats