Reducing the gap between Consumer and Retailer using Association Rule Mining and Classification Rule: An Overview
0 likes28 views
This document discusses how data mining techniques like association rule mining and classification rules can be used to analyze customer purchase data and reduce the gap between consumers and retailers. It proposes applying association rule mining to identify relationships between frequently purchased products, and using classification rules to categorize consumers by attributes like age, income, and marital status. This combined approach could provide retailers with useful insights into purchasing trends and patterns to help them better understand customer needs and improve marketing decisions.
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 11 | Nov -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 163
Reducing the gap between Consumer and Retailer using Association Rule
Mining and Classification Rule: An Overview
Arpit Kalra1, Harsh Mehta2, Om laddha3
1,2,3 SVKM’s NMIMS UNIVERSITY, Shirpur
------------------------------------------------------------------------***-----------------------------------------------------------------------
Abstract: In this study, we focused on how data mining can
be applied in Market Basket Analysis to identify new trend
sand purchasing patterns of customers. Data Mining is the
process of extracting useful information from a large
dataset. We are here using Association Rule to identify the
relationship between different products and Classification
Rule on the consumers to distinguish them on the basis of
pre-defined parameters. The valuable information
discovered from collaboration of these two algorithms
supports the retailers in decision making and hence increase
their sales.
I. INTRODUCTION
In today’s world large amount of data is generated every
day and this data is maintained in database in various
fields such as healthcare, education, market basket
analysis, etc. With this increasing data size, there is a need
to understand large and complex data and draw necessary
conclusions. The practice of extracting useful information
from large pre-existing databases is called Data Mining. It
is getting difficult for the local retailers to attract
customers, so there is a need for the retailers to
understand the shopping trends of the customers.
Many consumers prefer online shopping. With the growth
of the e-commerce websites, retailers tend to fail to attract
more and more consumers.
And this problem can be resolved by applying data mining
techniques to analyze new patterns and trends. We will
apply data mining techniques to the gathered data of the
customer behavior pattern, so that retailers will able to
know the new patterns and trends.
A. STEPS INVOLVED IN DATA MINING
a. Identifying the source information
b. Picking the data points that need to be analyzed
c. Extracting the relevant information from the data
d. Identifying the key values from the extracted data
set
e. Interpreting and reporting the results
II.DATA MINING TECHNIQUES
A. Association Rule
This rule is used to establish a relationship between
different objects that exist in the market. [1] This rule
is helpful in Market Basket Analysis.
B. Classification
This rule is used to classify a data item into predefined
classes. Foreg, we can use this rule to simply differentiate
cars into various categories. (Sedan, SUV)
Same principles can be applied to consumers, for e.g. by
distinguishing them into income, age and social group.
C. Clustering
With this technique, data is organized and classified into
meaningful subgroups or clusters.
D. Prediction
This technique is used to predict new information from a
set of existing data. For example Sales in future week can
be predicted using this rule.](https://image.slidesharecdn.com/irjet-v4i1128-171207090152/85/Reducing-the-gap-between-Consumer-and-Retailer-using-Association-Rule-Mining-and-Classification-Rule-An-Overview-1-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 11 | Nov -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 164
E .Outlier Analysis
This technique identifies and explains exceptions. For e.g.,
in Market Basket Data Analysis, Outlier can be some
transactions which happen uncommon.
III. EXISITING ALGORITHM
There are many algorithms used in Market Basket Analysis
for identifying the changes in purchasing trends of
customers. Some algorithms only focused on the
customers while others focused only on the products.
IV. PROPOSED METHOD
We are proposing in this paper to collaborate Association
and Classification Rule which will focus on both the
products and the customer’s .Association Rule will be
implied on the products and Classification Rule will be
implied on the customers. So this will be more effective for
the retailers to determine the purchasing trends of
customers.
V.ASSOCIATION RULE
Association rule is useful for making a simple relationship
between two or more items. This rule can be used for
tracing customers buying habits.
We might observe that a consumer always purchases socks
when he purchases shoes, & therefore suggest that in
future when he purchases shoes he might also want to
purchase socks.
Example:
Here we identify an important relationship between Milk
and butter. Association rule shows us that whenever a
customer buys Milk he always buys butter along with it.
Milk->Butter
This rule is helpful for retailers to identify the purchasing
patterns of the customers. And they will hence come to
know about what the customer wants.
Understanding Association rule with example
This is a dataset containing items purchased from a
retailer.
Tr.ID Product 1 Product 2 Product 3
1 Bread Butter Jam
2 Bread Butter Jam
3 Bread Butter Jam
4 Bread Butter
5 Bread Butter
Tr.ID-Transaction ID
For the above dataset we can establish following relations
between the different items:
Rule no 1: IFBread is bought, THEN butter is also
bought.
Rule no 2: IF butter is bought, THENBread is also
bought.
Rule no 3: IFBread and butter is bought, THENJam is
also bought in 60% of the transactions.
“IF THEN” format is followed in Association Rule.
The following terminologies defined are adapted from
Data Mining for Business Intelligence, by GalitShmueli and
others[4].
Frequent ProductSet :– Product set that is purchased
most number of times. For e.g., in the above dataset,
Bread&butterappeared in 100% of the purchases made.
Support :-This rule signifies the effect in terms of overall
size. If only a fewnumber of transactions are affected, the
rule may be of less use. For e.g, the support of “IF
BreadandButter THEN Jam” is 3/5 i.e. 60% of the total
transactions.
Confidence:- This determines functional use of the rule.
Transactions having confidence greater than 50% are
selected. For e.g., confidence of Bread, butter and Jam
given can be written as:](https://image.slidesharecdn.com/irjet-v4i1128-171207090152/85/Reducing-the-gap-between-Consumer-and-Retailer-using-Association-Rule-Mining-and-Classification-Rule-An-Overview-2-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 11 | Nov -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 165
The transactions containingBread&Butter (Antecedent)
and Jam (Consequent) are 3
The transactions containing only Bread&Butter
(Antecedent) are 5.
P(Bread&ButterandJam)/P (BreadandButter) = 3/5 =
60%
Thus we can conclude that the association rule is having a
confidence of 60%. More the confidence, Stronger the rule
is.
VI. CLASSIFICATION RULE
We can use classification rule mining to differentiate
customers or items on the basis of different parameters.
Here we are using Classification rule to classify the
customers by age, income, frequency of visits, marital
status, etc.
Classification rule helps the retailers to take decisions
easily.
Age Frequency Marital Status Frequency
Below 20 30 Single 45
20-40 50 Married 45
Above 40 40
Income Frequency Gender Frequency
<1Lakh 20 Male 70
1Lakh-5Lakh 45 Female 50
>5Lakh 25
The motivation for applying data mining approach on
Market Basket Analysis is to learn about buying patterns
and retailers can use this information so more no of
consumers are attracted towards them.
VIII. REFERENCES
[1]ManpreetKaur ,Shivani Kang*,Market Basket Analysis:
Identify the changing trends of market data using
association rule mining,
[2]NeşeAcar, BülentÇizmeci, Factors Influencing
Customer’s Choice of Technology Retailers: An Application
In Kayseri (Turkey),
[3]NeeshaJothiNur’Aini,AbdulRashidWahidahHusain, Data
Mining in Healthcare – A Review,
[4]https://www.analyticsvidhya.com/blog/2014/08/effec
tive-cross-selling-market-basket-analysis/
Retailers will have the knowledge about the customers
according to these defined parameters, in this case age,
marital status, income, gender.
Suppose in an area if there are more number of students
(age <20) so the retailer will focus more towards the needs
of the student like stationary items and sports items.
VII. CONCLUSION
Data Mining has played a very important role in Market
Analysis and various other fields. The most important
point to succeed in a marketing strategy is to create an
accurate customer analysis[2].](https://image.slidesharecdn.com/irjet-v4i1128-171207090152/85/Reducing-the-gap-between-Consumer-and-Retailer-using-Association-Rule-Mining-and-Classification-Rule-An-Overview-3-320.jpg)
Reducing the gap between Consumer and Retailer using Association Rule Mining and Classification Rule: An Overview
- 1. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 04 Issue: 11 | Nov -2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 163 Reducing the gap between Consumer and Retailer using Association Rule Mining and Classification Rule: An Overview Arpit Kalra1, Harsh Mehta2, Om laddha3 1,2,3 SVKM’s NMIMS UNIVERSITY, Shirpur ------------------------------------------------------------------------***----------------------------------------------------------------------- Abstract: In this study, we focused on how data mining can be applied in Market Basket Analysis to identify new trend sand purchasing patterns of customers. Data Mining is the process of extracting useful information from a large dataset. We are here using Association Rule to identify the relationship between different products and Classification Rule on the consumers to distinguish them on the basis of pre-defined parameters. The valuable information discovered from collaboration of these two algorithms supports the retailers in decision making and hence increase their sales. I. INTRODUCTION In today’s world large amount of data is generated every day and this data is maintained in database in various fields such as healthcare, education, market basket analysis, etc. With this increasing data size, there is a need to understand large and complex data and draw necessary conclusions. The practice of extracting useful information from large pre-existing databases is called Data Mining. It is getting difficult for the local retailers to attract customers, so there is a need for the retailers to understand the shopping trends of the customers. Many consumers prefer online shopping. With the growth of the e-commerce websites, retailers tend to fail to attract more and more consumers. And this problem can be resolved by applying data mining techniques to analyze new patterns and trends. We will apply data mining techniques to the gathered data of the customer behavior pattern, so that retailers will able to know the new patterns and trends. A. STEPS INVOLVED IN DATA MINING a. Identifying the source information b. Picking the data points that need to be analyzed c. Extracting the relevant information from the data d. Identifying the key values from the extracted data set e. Interpreting and reporting the results II.DATA MINING TECHNIQUES A. Association Rule This rule is used to establish a relationship between different objects that exist in the market. [1] This rule is helpful in Market Basket Analysis. B. Classification This rule is used to classify a data item into predefined classes. Foreg, we can use this rule to simply differentiate cars into various categories. (Sedan, SUV) Same principles can be applied to consumers, for e.g. by distinguishing them into income, age and social group. C. Clustering With this technique, data is organized and classified into meaningful subgroups or clusters. D. Prediction This technique is used to predict new information from a set of existing data. For example Sales in future week can be predicted using this rule.
- 2. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 04 Issue: 11 | Nov -2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 164 E .Outlier Analysis This technique identifies and explains exceptions. For e.g., in Market Basket Data Analysis, Outlier can be some transactions which happen uncommon. III. EXISITING ALGORITHM There are many algorithms used in Market Basket Analysis for identifying the changes in purchasing trends of customers. Some algorithms only focused on the customers while others focused only on the products. IV. PROPOSED METHOD We are proposing in this paper to collaborate Association and Classification Rule which will focus on both the products and the customer’s .Association Rule will be implied on the products and Classification Rule will be implied on the customers. So this will be more effective for the retailers to determine the purchasing trends of customers. V.ASSOCIATION RULE Association rule is useful for making a simple relationship between two or more items. This rule can be used for tracing customers buying habits. We might observe that a consumer always purchases socks when he purchases shoes, & therefore suggest that in future when he purchases shoes he might also want to purchase socks. Example: Here we identify an important relationship between Milk and butter. Association rule shows us that whenever a customer buys Milk he always buys butter along with it. Milk->Butter This rule is helpful for retailers to identify the purchasing patterns of the customers. And they will hence come to know about what the customer wants. Understanding Association rule with example This is a dataset containing items purchased from a retailer. Tr.ID Product 1 Product 2 Product 3 1 Bread Butter Jam 2 Bread Butter Jam 3 Bread Butter Jam 4 Bread Butter 5 Bread Butter Tr.ID-Transaction ID For the above dataset we can establish following relations between the different items: Rule no 1: IFBread is bought, THEN butter is also bought. Rule no 2: IF butter is bought, THENBread is also bought. Rule no 3: IFBread and butter is bought, THENJam is also bought in 60% of the transactions. “IF THEN” format is followed in Association Rule. The following terminologies defined are adapted from Data Mining for Business Intelligence, by GalitShmueli and others[4]. Frequent ProductSet :– Product set that is purchased most number of times. For e.g., in the above dataset, Bread&butterappeared in 100% of the purchases made. Support :-This rule signifies the effect in terms of overall size. If only a fewnumber of transactions are affected, the rule may be of less use. For e.g, the support of “IF BreadandButter THEN Jam” is 3/5 i.e. 60% of the total transactions. Confidence:- This determines functional use of the rule. Transactions having confidence greater than 50% are selected. For e.g., confidence of Bread, butter and Jam given can be written as:
- 3. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 04 Issue: 11 | Nov -2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 165 The transactions containingBread&Butter (Antecedent) and Jam (Consequent) are 3 The transactions containing only Bread&Butter (Antecedent) are 5. P(Bread&ButterandJam)/P (BreadandButter) = 3/5 = 60% Thus we can conclude that the association rule is having a confidence of 60%. More the confidence, Stronger the rule is. VI. CLASSIFICATION RULE We can use classification rule mining to differentiate customers or items on the basis of different parameters. Here we are using Classification rule to classify the customers by age, income, frequency of visits, marital status, etc. Classification rule helps the retailers to take decisions easily. Age Frequency Marital Status Frequency Below 20 30 Single 45 20-40 50 Married 45 Above 40 40 Income Frequency Gender Frequency <1Lakh 20 Male 70 1Lakh-5Lakh 45 Female 50 >5Lakh 25 The motivation for applying data mining approach on Market Basket Analysis is to learn about buying patterns and retailers can use this information so more no of consumers are attracted towards them. VIII. REFERENCES [1]ManpreetKaur ,Shivani Kang*,Market Basket Analysis: Identify the changing trends of market data using association rule mining, [2]NeşeAcar, BülentÇizmeci, Factors Influencing Customer’s Choice of Technology Retailers: An Application In Kayseri (Turkey), [3]NeeshaJothiNur’Aini,AbdulRashidWahidahHusain, Data Mining in Healthcare – A Review, [4]https://www.analyticsvidhya.com/blog/2014/08/effec tive-cross-selling-market-basket-analysis/ Retailers will have the knowledge about the customers according to these defined parameters, in this case age, marital status, income, gender. Suppose in an area if there are more number of students (age <20) so the retailer will focus more towards the needs of the student like stationary items and sports items. VII. CONCLUSION Data Mining has played a very important role in Market Analysis and various other fields. The most important point to succeed in a marketing strategy is to create an accurate customer analysis[2].