laporan ar mba ip 4

7/23/2019 Laporan ar mba ip 4

http://slidepdf.com/reader/full/laporan-ar-mba-ip-4 1/23

CHAPTER I

INTRODUCTION

1.1 Description of the Problem

The development and comptetition in world trading throught free market economy and

information technology advance bring company in more opening tight competition level

to fulfill high customer expectation. The company must be apply good business standard

if they still want to compete and still has a market share. The competition in this business

cannot be separated from information technology that become the current issue today.

In several years, the development of retail business has a significant improvement,

especially in Yogyakarta. Every years many people move to Yogya because of their job

and business or the student who want to continue their study here. Based from that

situation this is the good chance to develop retail business in Yogyakarta. This is because

everyday people need to fulfill their daily needs, so the demand for retail industry in

Yogyakarta is still high. However their faced a new problem now. The new problem is

with recent market growing now, the competition between retail shop are tighter than

before. One of the example from competition is between Swalayan Kopma UGM and

Mirota Kampus that has closed location . Both of them are big retail store and has many

customer who visited and purchased their selling item everyday.

Swalayan Kopma UGM are located at Jalan Kaliurang Bulaksumur H-8 in front of

Vocational School Gadjah Mada University. Swalayan Kopma UGM is the retail with

supermarket format. Swalayan Kopma UGM are directed to become primary division

that will be provide basic needs for Gadjah Mada University student, such as college

equipment like stationary, tabloid, bulletin, journal, and magazine. They also have

grocery item to fulfill customer daily needs and Gadjahmada University accessories like

stickers, callendar, etc. Development of Swalayan Kopma UGM will be directed into

student shop center (Kopma UGM, 2015).

However, because of the competition with another retail shop they need to make a

strategy to defend in this retail business. From that a supermarket must understand what

their customer wants and needs to make them comfortable while shopping in that shop,



especially to make customer choose their needed item easily. For the example, in placing

their shopping item that are composed in rack should fit to customer shopping pattern.

The problem that Swalayan Kopma UGM faced now is the layout of item placement

still not fit with the customer shopping pattern. This is because they still don’t see their

customer pattern who will purchased a product which placed near to each other and will

be purchase together. Based from that, very important to placed item that are suitable

with customer consumption pattern that actually can influence customer shopping tastes

and the selling for that product (Albion Research, 2007 in Lestari, 2009). The placement

item in a rack from a supermarket can be analyze from product selling transaction data ,

however Swalayan Kopma UGM still not using that transaction data to process become

knowledge that will be improve their business profit with the optimal placement of the

item.

One of the placement method that can analyze customer shopping behavior pattern is

Market Basket Analysis (MBA). This method is one of the method in data mining that

purpose to find the most selling products togeher based from transaction data. Analysis

pattern method of MBA shopping behavior used apriori algorithm, which will be used to

get association rule with pattern “if then”. That technique can be apply in very big datalike selling transaction data (Marsela dkk, 2004 in Bonai, 2011). With using data mining

technique will help people to can use MBA without using manual method.

Based from the explanation above, the objective of this research is design a new

layout solution for Swalayan Kopma UGM.

1.2

Problem Formulation

Problem formulation in this module MUST consists of:

1. How is the associative relations that happened between items in the Swalayan

Kopma UGM?

2.

How is the layout solutions for Swalayan Kopma UGM based on analysis of AR-

MBA?



CHAPTER II

LITERATURE REVIEW

2.1. Deductive Study

2.1.1. Market Basket Analysis

Market Basket Analysis is a method or the often and most helpful technique for marketing

environment, this technique used association techniques from data mining. Association

techniques is a data mining technique for finding associative rules between combination

item (Eibe Frank et al, 2011). The objuective from this method is for determine which product will be bought together by customer at the same time.

Like an example how most likely a customer will buy bread together with milk. With

that knowledge, the owners of retail store or company can used that information to setting

a item layout so the most selling item together will be placed together in one area. Then,

the company that sell item online can used that information to setting a layout in their

store.

2.1.2. Association Rules

Association analysis or association rule is the technique in data mining to finding

associative rules between an item (Lutfi, 2009). Association rule has two steps, first search

combination which often happened in one item, then defining condition and result.

Like an example, associative rules from purchased analysis in a retail store is we can

know how big the possibility a customer will purchased bread together with milk. With

that knowledge, the owner of retail store can set the item layout or design market

campaign with using discount coupon for selected combination item.

The association rule that usually stated in:

{Bread, Margarine}→ {Milk} (Support = 40%, confidence =50%)

That rule means 50% from transaction in database which purchased bread and

margarine will purchased milk also. When 40% from all of the transaction in database has

three of that item. That also mean one customer that purchased bread and margarine has

50% possibility for purchased milk. This is significant enough because that represent

40% from all of transaction notes.

The customer behavior to implemented association analysis also need to understand if



customer took one or batch of first item, this can be said a condition or left hand. Item or

batch of item that will be purchase next is result or right hand. For make an effective rule

in market basket analysis, there are three size that must be note such as “Support Score”,

“Confidence Score”, and “Improvement” or “Lift Score”.

a. Support

Support is a size that show percentage from basket where left hand and right hand both of

them are found together (Marakas, 2003). We can said that support is a buyer percentage

support who purchased item condition will purchase result item also from all of

transaction.

From that statement can be write as:

( Re) = (+)

x 100%

Formula for single selling item can be write as:

() =()

x 100%

The character of support is bidirectional, means support (A+B) is equal to support

(B+A).

b.

Confidence

This size are different from support because confidence is the possibility of apperance of

right hand item with knowing that left hand item already in shopping basket (Markas,

2003).

From the statement above the formulation is:

= (+)

() x 100%

Confidence won’t has many meaning if the item which is not sold are input in one

basket together with sold item. The characteristic of confidence is undirectional, so

confidence (A+B) are not definitely equal to confidence (B+A).

c. Improvement/Lift

After knowed the support that has bidirectional character and confidence that has

undirectional character, we want to know also is that true both of the rules are valid or

invalid. Improvement or lift is the number from combination of confidence divided with

support result, so the formula is:

= (+)

() x 100%



The good rule is the rule that has improvement score more than 1,0 (Edi Suryanto, et

al. 2015).

2.2. Inductive Study

Research conducted by Heru Dewantara et al. (2013) with the title "Designing

Applications Data Mining with the Apriori Algorithm to Frequency Market Basket

Analysis on Sales Transaction Data”, researcher trying to made business strategy in

layout shopping item that adjusted to customer consumption pattern in Swalayan KPRI

Brawijaya University Malang. The method that researcher used is Market Basket

Analysis (MBA) with using transaction selling data during February 2013. The result of

this research is the MBA application prototype. Examination prototype did with

transaction minimum limitation (support) are 7 transaction an cofidence minimum are

5%. With that limitation, there are 11 association rules from MBA application. One of

association rule is if the customer purchased 1 kg sugar, Indofood Bumbu Racik Sayur

Sop 20gr, the customer will purchased Indofood Bumbu Racik Sayur Asem 20gr with

support score = 0,52% and confidence score 90,91% which will be highes confidence

score. Next process is categorized the used item as reference for making a solution

layout, so the conclusion is sugar will be placed near egg, instant seasoning, and oil; oil

will be placed near instant seasoning; egg will be placed near to rice and noodle also

beverage will be placed near to bread. With that, the layout of selling item can be adapted

with association rule and compatible with cusomer consumption pattern.

Research conducted by Olivia (2015) with the title "Perancangan Sistem Informasi

Data Mining Dengan Algoritma Apriori untuk Penentuan Layout Produk Pada PT. Metro

Makmur Nusantara”. PT. Merto Makmur Nusantara (Metro Supermarket) is one of many

recent growing retail business in Medan and the retail that oriented in food product and

non – food product. PT. Metro Makmur Nusantara already used Magic Retail

Information System (MARIS) to saves and process all of the selling data in company.

However, the program they applied cannot give an information that relationship with

taking a decision like market basket analysis from customer.



To did that, they need to apply data mining method. Data mining is information

extract or important or interesting pattern from data in the unknown big base data, but

has a useful potential information. Apriopri Algorithm is the collected data algorithm

with association rule to determine an item combination association relationship.

Association Rule did with calculation mechanism support and confidence from an item

relationship. The result of this result is a data mining application that support company

management to know the most often purchased item together in one purchasing

transaction so the researcher can setting product layout. This will improved time

efficiency for customer to search and purchased item in supermarket.



CHAPTER III

RESEARCH METHOD

3.1 Object of the Research

The object in this Association Rule – Market Basket Analysis research is in Swalayan

Kopma UGM at Jl. Kaliurang, H-8 Bulaksumur, Yogyakarta.

3.2 Collecting Data Method

The method of this research are took a group of sample that can represent population

with collected minimum 100 receipts, where the receipt are fulfill data pre-processing

condition without used failed transanction, not used one item only transaction data, and

also not use one department in one transaction data only. Researcher expected sample

which will be used in AR – MBA will be represent existing population, because this

analysis become good if sample are representative. We believe representative sample to

this population also can represent population.

3.3 Types of Data

3.3.1. Primary Data

Primary data in this research are customer shopping receipts which are collected through

collective process in location directly.

3.3.2. Secondary Data

Secondary data in this research is data that already proceed from primary data to data that

ready to further process. Secondary data in this research are buying data, transformation

data, and tabulation data and then ready to process with using software.



3.4 Flowchart

Process to did this research are illustrated in flowchart below:

START

INPUT

DATA

Pre Processing Data:

1. Integration Data

2. Transformation Data

Tabulation Data

Association Rule

Solution Layout

FINISH

Figure 3.1 Flowchart AR – MBA



The explanation in flowchart above:

a. Start the software, there are two software that researcher used in this report. There are

Microsoft Excel and Rapid Miner. The first step, start Microsoft Excel.

b. First input data recapitulation in Microsoft Excel. Data that will input in Microsoft

Excel are the item that customer purchased based from their receipt. Minimum item to

did this research are 2 different department in one receipt.

c.

After all of the data are recapitulated, the next process is integrate data from each

category. First, divided each item in several department. For determination of each

department are subjective, based from every researcher preferences to placed them in

each department. . When finished to integrated the data, after that transform data to become compatible form. After decided department for each item, copy the recapitulation

in the next sheet. Next click find in excel, click name of item, replace the name of the

item with typed the name of the department in that item, finally click replace all. Repeat

that step until one department are finished and repeat again to the next department until

all of department are finised. If there are 2 or more item in one department in one

transaction id, just choose one of them and delete the item in the same department.

d.

Process all of data become tabulation data, researcher need to change data department

to matrix binary. Binary number is the two character number 0 and 1. The reason to

transform the data is because Rapid Miner software just can process the output if the

input are made in matrix binary. First made a table that consist of departement and

transaction id. Next, give the number 1 for every department that are purchased in that

transaction id, and 0 if that department aren’t purchased. Finally the file are ready to

process using Rapid Miner software.

e.

Then, used Rapid Miner Software to got Association Rule and FP - Growth output.

First, open Rapid Miner Software. After that click new process. Then, click file, choose

import data, and choose import excel sheet. Next, search our excel data in computer then

click next. After choose the data, click next. The excel file will be opened in data import

wizard window choose matrix binary data and click next. In step three there is a preview

of tabulation data without name, just click next. For step 4 make sure all of box already

check, click next, save the data, and click finish.



After the output are released when click finish, click main process. Drag file in main

process box. Search select attributes, then drag to main process box. Connect file and

select attributes, click select attributes and replace attribute filter by with subset, then

click select attributes menu in attributes, move all of department in right box. Search

numerical by binomial and remap binomials, then drag to main process box. Connect

select attributes with numerical by binomial and numerical to binomial with remap

binomial.After that replace negative value with 0 and positive value with 1. Search FP –

Growth and create association rules, then drag to main process box, after that connect

remap binomials to FP Growth and FP – Growth (fre) to create association rules, next

connect FP – Growth (exa) to first res and create association rules (rul) to second res.

After that, click FP – Growth fill the value of min support and click create association

rule, then replace minimum confidence value. The value both of min support and min

confidence are subjective based from researcher preference. Finally, click Run and the

output of the process will be released. The association rule output is possibility if the

item in one department are purchased together. FP Growth output is the most selling

item.

f. Finally, based from association rule and fp growth output we can read the conclusion

of both output. After that, made a solution layout for this case. For example, the most

selling item must be placed near to cashier and placed together with the most possibility

item that will be purchased together with the most selling item. Researcher using

Microsoft Visio to made a layout solution in this case.



CHAPTER IV

RESULT AND DISCUSSION

4.1. Initial condition of the research object

Swalayan Kopma UGM is the retail shop that has many visitor enough everyday,

especially in school day when student has class in their university. Because most of

this shop customer are student, most of the selling item are for student needs, like

stationary and item for student of Gadjahmada University. Then this shop has many

item besides stationary and UGM student item like tabloid, magazine, newspaper, and

even daily needs.

The case study now are about layout in Swalayan Kopma UGM. Researcher went

to Swalayan Kopma UGM to collected shopping receipt. After that researcher analyze

if the layout already compatible or not based from what the most purchased item by

customer. From all of that item they are selling, the layout of Swalayan Kopma UGM

like the picture below.



Figure 4.1 Initial Layout of Swalayan Kopma UGM

In this layout, the item which placed closed to cashier are candies, batteries and

bread. After that shampoo, cosmetic, and cleanser are placed closed each other. Snack,

instant noodle, and household equipment are placed closed each other. Beverages are

placed close to raincoat. Raincoat is placed close to clothes and accessories. Sachet

drink, biscuit and instant seasoning are placed close to each other.

4.2. Rapid Miner Output

After collected all of the receipt, researcher need to recapitulated it and made several

department based from all of item which customer bought. Then, made a

preprocessing data and next made transformation data in Microsoft Excel. Finally,

input excel file to software Rapid Miner, and the output will be calculated like picture

and table below.



Figure 4.2 Main Process

After made main process, click run and the output will be shown like table below.

FP – Growth is the table that shown the most purchased item by customer.

Table 4.1 FP - Growth

Size Support Item 1 Item 2

1 0.630 DEPT.12

1 0.200 DEPT.6

1 0.190 DEPT.23

1 0.170 DEPT.7

1 0.150 DEPT.14

1 0.140 DEPT.25

1 0.140 DEPT.16

1 0.140 DEPT.13

1 0.130 DEPT.17

1 0.130 DEPT.11

1 0.100 DEPT.18



Size Support Item 1 Item 2

1 0.090 DEPT.4

1 0.080 DEPT.5

1 0.080 DEPT.22

1 0.070 DEPT.9

1 0.050 DEPT.3

1 0.050 DEPT.20

2 0.140 DEPT.12 DEPT.6

2 0.070 DEPT.12 DEPT.23

2 0.070 DEPT.12 DEPT.7

2 0.080 DEPT.12 DEPT.14

2 0.080 DEPT.12 DEPT.25

2 0.080 DEPT.12 DEPT.16

2 0.060 DEPT.12 DEPT.13

2 0.070 DEPT.12 DEPT.11

2 0.060 DEPT.12 DEPT.4

2 0.080 DEPT.12 DEPT.5

Association Rules is table that describe if “Item B” will be choose if “Item A” are

bought. The table of association rules are like the table below.

Table 4.2 Association Rules

No Premises Conclusion Support Confidence La - Place Gain p-s Lift Confiction

1 DEPT.12 DEPT.23 0.07 0.11 0.656 -1.19 -0.049 0.584 0.911

2 DEPT.12 DEPT.7 0.07 0.11 0.656 -1.19 -0.037 0.653 0.933

3 DEPT.12 DEPT.11 0.07 0.11 0.656 -1.19 -0.012 0.854 0.978

4 DEPT.12 DEPT.14 0.08 0.126 0.662 -1.18 -0.015 0.846 0.973

5 DEPT.12 DEPT.25 0.08 0.126 0.662 -1.18 -0.008 0.907 0.985

6 DEPT.12 DEPT.16 0.08 0.126 0.662 -1.18 -0.008 0.907 0.985

7 DEPT.12 DEPT.5 0.08 0.126 0.662 -1.18 0.0296 1.587 1.053

8 DEPT.12 DEPT.6 0.14 0.222 0.699 -1.12 0.0140 1.111 1.028

9 DEPT.23 DEPT.12 0.07 0.368 0.899 -0.31 -0.049 0.584 0.585

10 DEPT.7 DEPT.12 0.07 0.411 0.914 -0.27 -0.037 0.653 0.629



No Premises Conclusion Support Confidence La - Place Gain p-s Lift Confiction

11 DEPT.13 DEPT.12 0.06 0.428 0.929 -0.220 -0.028 0.680 0.647

12 DEPT.14 DEPT.12 0.08 0.533 0.939 -0.219 -0.014 0.846 0.792

13 DEPT.11 DEPT.12 0.07 0.538 0.946 -0.19 -0.012 0.854 0.801

14 DEPT.25 DEPT.12 0.08 0.571 0.947 -0.2 -0.008 0.907 0.863

15 DEPT.16 DEPT.12 0.08 0.571 0.947 -0.2 -0.008 0.907 0.863

16 DEPT.4 DEPT.12 0.06 0.666 0.972 -0.12 0.0033 1.058 1.11

17 DEPT.6 DEPT.12 0.14 0.7 0.950 -0.26 0.0140 1.111 1.233

18 DEPT.5 DEPT.12 0.08 1.0 1.0 -0.08 0.0296 1.587 Infinity

4.3.

Analysis of the data processing result from the point of view of its consumer

behavior

From the data that already got from 100 receipt in Swalayan Kopma UGM Jl.

Kaliurang Bulaksumur, researcher did a researched to know asosiation relationship

between item that related to customer behavior. The item which bought by the

customer are the heterogen item that classified to be several department. After

processed the data, there are 25 department such as newspaper, staple food, sugar,

snack, bread, biscuit, candy, ice cream, instant noodle, spice, drink sachet, beverages,

sanitary napkin and tissue, clothes and accessories, shoe polish stationary, cosmetics,

tooth brush and paste, shampoo, soap, cleaner, household appliance, medicine, baby

equipment, and ciggarette and crickets.

Based from calculation using Rapid Miner software, the result in Frequent Pattern

Growth ( F-P Growth) table are Department 12(beverage) has 0,63 support,

department 6 (biscuit) has 0,20 support, department 23(medicine) has 0,19 support

,department 7 (candies) has 0,17 support, department 14 (clothes and accessories) has0,15 support score, department 25 (ciggerattes and crickets) has 0,14 support score,

department 16 (stationary) has 0,14 support, department 13 (sanitary napkin and

tissue) has 0,14, department 17 (cosmeticts) has 0,13 support score, department 11

(sachet drink) has 0,13 support, department 18 (tooth brush and paste) has 0,10

support, department 4 (snack) has 0,09 support, department 5 (bread) has 0,08

support, department 22 (houshold appliaces) has 0,08 support, departement 9 (instant

noodle) has 0,07 support, department 3 (sugar) has 0,05 support, and department 20(soap) has 0,05 support. If there are 2 item which frequent be choosen by the



customer, there are department 12 and department 6 which has 0,14 support,

department 12 and department 23 has 0,07 support, department 12 and department 7

has 0,07 support, department 12 and department 14 has 0,08 support, department 12

and 25 has 0,08 support, department 12 and department 16 has 0,08 support,

department 12 and department 13 has 0,06 support, department 12 and department 11

has 0,07 support, department 12 and department 4 has 0,06 support, and department

12 and 5 has 0,08 support. In FP- Growth table result, researcher used 0,2 as the

minimum support for this process.

For Association Rule table, researcher used 0,1% minimum confident. The result

of the association rule are:

a.

Rule 1

If customer purchased item in department 12, customer must be purchased item in

department 23 with 0,07 support score and 0,11 confident score.

b. Rule 2



c. Rule 3



d. Rule 4



e. Rule 5



f.

Rule 6



g.

Rule 7





h.

Rule 8



i.

Rule 9



j. Rule 10



k. Rule 11



l. Rule 12



m. Rule 13



n.

Rule 14



o.

Rule 15



p.

Rule 16



q. Rule 17



r. Rule 18


department 12 with 0,08 support score and 1 confident score.



From the result both of FP – Growth and Association rule, the conclusion are the

most purchase item by cutomers are the item from department 12 (beverages), and in

association rule all of item from each department must be pairing with item in 12

department when customer shopping in Swalayan Kopma UGM.

4.4. Solution recommendations based on the associations rule result

After found result of the Frequent Pattern Growth (FP – Growth) and Association rule,

researcher can found the solution layout where:

a.

The most purchased item (department 12) must be close to cashier. This is

because cashier are closed to door, so this will make customer take this item easier

than to placed it far enough from cashier like the initial layout.

b. The department which always choose after department 12 must be placed closed

to department 12. This will make customer can search and take the item easily. From

that result, the item that must be placed closed department 12 are department 23,

department 7, department 11, department 14, department 25, department 16,

department 5, and department 6.

c. The item which placed near to cashier but not absolutely need to placed close to

department 12 are department 4, department 13, department 17, department 18,

department 4, department 22, department 9, department 3, and department 20.

Based from that result, the layout solution that researcher recommend to Swalayan

Kopma UGM, are like the figure below.



Figure 4.3 Solution Layout for Swalayan Kopma UGM

In this solution layout, researcher purposely made the layout like this based from

the result of FP – Growth and Association Rules. Researcher hope with this new

layout, customer can search the item easily especially for the item which are

purchased together with the most selling item. Then the item which are placed near to

cashier can be found easily by the customer. Finally, because of that simplicity can

improve customer satisfaction, and improve business profit.



CHAPTER V

CONCLUSION AND RECOMMENDATION

5.1 Conclusion

From all result of calculation in Rapid Miner software, researcher conclude that

a. From FP – Growth result the most selling item are from department 12 (beverages).

Then in Association Rule output, when customer purchased item in department 12, they

must be purchased item from department 23, department 7, department 11, department

14, department 25, department 16, department 5, and department 6.

b.

Based from the result of the Association Rule and FP - Growth, the layout solution for

this case are researcher placed the most selling item near to cashier and the item that will

be purchased together with most selling item will be placed closed to each other.

5.2 Recommendation

5.2.1.

For Future Research

Recommendation for future research, based from research limitation. Limitation in this

research

a. Limitation of the population sample, researcher just took 100 sample receipt in

Swalayan Kopma UGM, in the future researcher recommend to take more sample so the

improve support value so the item appear frequently.

b. Limitation of collecting data method, researcher just took a receipt to made this report.

For the future researcher recommend to add collecting data method, like interview to the

customer or the expert in Swalayan Kopma UGM.

5.2.1.

For Swalayan Kopma UGM

Researcher recommend Swalayan Kopma UGM to repair their initial layout with placed

the most selling item near to the cashier followed by the item which are always

purchased together with the most selling item.



REFERENCE

Dewantara, Heru et al. 2013. Designing applications data mining with the apriori

algorithm to frequency market basket analysis on sales transaction data

(Case Study in Supermarket KPRI University of Brawijaya).Malang:

University of Brawijaya.

Olivia. 2015. Perancangan sistem informasi data mining dengan algoritma apriori

untuk penentuan layout produk pada pt. Metro makmur nusantara. Medan:

STMIK TIME

Suryanto, Edi et al. Implementasi customer relationship management dengan

market basket analysis pada toko buku online (studi kasus: toko

buku toga mas).

Kopma UGM. 2015. Divisi usaha. (online): http:kopmaugm.com/divisi-usaha(June 21st 2015 )



ATTACHMENT



ANALYSIS DECISION AND DATA MINING ASSOCIATION RULE–MARKET BASKET ANALYSIS (AR-MBA)

IP – 4 GROUP 4

1.

Syarifah Nabila AR

ASSISTANT DM - 40

DATA MINING LABORATORY

INDUSTRIAL ENGINEERING DEPARTMENT

INDUSTRIAL TECHNOLOGY FACULTY

ISLAMIC UNIVERSITY OF INDONESIA

2015

Scoring Criteria Max

Format 10

Content 50

Analysis 40

TOTAL 100

Yogyakarta,June 23rd 2015

Assistant

(Doni Hikmawan)

PRACTICUM REPORT

laporan ar mba ip 4

Documents