p–ISSN: 2723 - 6609 e-ISSN: 2745-5254

Vol. 5, No. 8 August 2024 http://jist.publikasiindonesia.id/

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3323

Clothing Product Recommendations Using the FP-Growth

Algorithm in Siny.CO Stores

Duny Muhamad Firmansyah

, Sunjana

Universitas Widyatama, Indonesia

Email:

[email protected]

, [email protected]

*Correspondence

ABSTRACT

Keywords: association;

clothing sales; data

mining; fp-growth;

rapidminer.

In this modern era, there is so much competition in the

business world, especially in the sales industry, that requires

shop owners to find a strategy that can increase sales and

marketing of the products they sell, one of which is by

utilizing clothing sales transaction data using data mining.

Data Mining is an iterative and interactive process to find

new patterns or models that can be generalized for the future,

useful and understandable in a very large database (massive

database). In these conditions, good data processing

techniques are needed, one of which is data mining

techniques. One thing that can be used in this technique is to

use the Association rule method using the FP-Growth

algorithm, which is an algorithm that produces frequent

itemsets which will later be used to determine

recommendations for clothing products for the needs of the

Siny.co Store. This research uses the RapidMiner Studio

application to help process transaction data. This research

method uses a minimum support of 20% and a minimum

confidence of 80%, thereby creating 9 association rules,

where these rules determine several products for

recommendation at the Siny.Co Store, including Sashi,

Canna, Alice, Tartan, Aruna, Cassandra, Nala, Lalita, and

Acio.

Introduction

The many competitions in the business world, especially in the sales industry,

require developers to find a strategy that can increase sales and marketing of the products

sold, one of which is the use of sales transaction data (Kumar & Dubey, 2023). With daily

sales activities, the data will increase over time. The data not only functions as an archive

for the company, it can also be used and processed into useful information for increasing

sales and product promotion (Allam & Dhunny, 2019).

From the source of clothing store sales data, it shows that the demand for clothing

is increasing. This is what is used as the basis for processing Data Mining in clothing

store sales. To manage this data, a method is needed that can be used to dig up information

from the data. This method is known as Data Mining. Data mining is an iterative and

Duny Muhamad Firmansyah, Sunjana

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3324

interactive process of finding new patterns or models in massive databases that are useful,

understandable, and generalizable for the future (Sari, Muhammad Syahril, Kom, &

Suharsil, 2021). Data mining involves searching large databases for trends and patterns

of interest in order to make future decisions (Saura, 2021). These patterns can be

recognized with specialized tools that can provide useful and insightful data analysis, and

can then be explored in more detail, perhaps using other decision support tools (Wardani,

2020).

Previous research on inventory systems using the FP-Growth Algorithm concluded

that data mining techniques with the FP-Growth Algorithm can be implemented in the

spice product inventory system (Kana, Ramadhan, & Mahyuni, 2022). With an

application based on information technology, a method that can increase knowledge to

provide goods by providing advice to companies on the relationship of consumers to an

item purchased by consumers can be calculated using the FP-Growth Algorithm

technique (Salu, Michael, Padang, & Adda, 2022).

The FP-Growth algorithm includes a type of association rule on Data Mining, the

FP-Growth Algorithm which aims to find frequent itemsets executed on a set of datam

(Wu & Zhang, 2023). FP-Growth analysis defines a process to find all FP-Growth rules

that meet the minimum requirements for support and the minimum requirements for

confidence (Shawkat, Badawi, El-ghamrawy, Arnous, & El-desoky, 2022). In this study,

the FP-Growth algorithm will be used for an association approach, so that the right

product recommendations will be found (Brous, Janssen, & Herder, 2020).

Digging association rules is a procedure to find relationships between items in a

dataset. Start by looking for frequent itemsets, which are the combinations that most often

occur in an itemset and must meet the minimum support (demand). Itemset mining that

often arises from large transactional databases is one of the most challenging problems in

data mining, in many real-world scenarios, data is not extracted from a single data source

but from distributed and heterogeneous data. The knowledge found is expected to help

better business operations (Merliani, Khoerida, Widiawati, Triana, & Subarkah, 2022). In

the data mining method, association rules are one of the most popular. However, digging

up information patterns that use association rules often results in very large individual

patterns, thus leaving the analyst to complete the task with all the rules and find the one

of interest (Hasan, Aziz, & Nofendri, 2023).

Based on research that has been conducted previously, including research entitled

"Application of the Apriori Algorithm in Sales Transactions for Food and Drink Menu

Recommendations". In this research, the Apriori method was used to find out the menu

at Warung Tenda to be used as a menu package with 50 sales transaction data for a total

of 10 items, 11 rule associations were formed which could use these results as reference

material to be added as a menu package to the Warung Tenda menu list (Hilman, 2022).

Clothing Product Recommendations Using the FP-Growth Algorithm in Siny.CO Stores

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3325

Research Methods

Figure 1. Research Methods

In conducting this research in order to get the appropriate results as desired, the

research method that has been determined in accordance with the algorithm in this study

is used. The following are the steps used in this research object using the FP-Growth

algorithm that has been adjusted, as shown in figure 1 above. The explanation of each of

the stages above is as follows:

Problem Identification

The problem identified in this study is to get recommendations related to clothing

products in the Siny.Co store to produce maximum product sales as well as maximize

product promotion by determining the best product recommendations.

Transaction Data Collection

The data collection stage is the second stage in doing this research task. The data of

this research was collected from transaction data on Siny.Co stores. The data comes from

transactions that have occurred, namely from January to November in 2022. In collecting

data using certain techniques or methods in the process. The following is an explanation

of data collection at the Siny.Co Shop:

study book

It is carried out with the aim of finding out what method will be used to solve the

problem to be researched, as well as obtaining a strong reference basis in applying a

method that will be used in this Research Project, namely by studying articles and journals

related to the problem of Clothing Product Recommendations in Siny.Co Stores using the

FP-Growth algorithm.

Observation

Duny Muhamad Firmansyah, Sunjana

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3326

The observation technique used is direct observation of the object being studied,

such as making direct observations at Siny.Co stores. Observation activities are carried

out in the circulation room to see transaction activities and products of goods sold.

Preprocessing Data

This stage includes the data processing process using a predetermined method,

namely the Association Rule Mining method using the FP - Growth Algorithm. The

process of processing sales transaction data is as follows:

1. Selection

At this stage, transaction data that contains information on the item number and

item code in the transaction data is not included in the dataset to be processed. In addition,

the price of purchased goods is not used in this study and is not included in the FP-Growth

process.

2. Cleaning

At this stage, the process of correcting incomplete or blank data is carried out

because the transaction is recorded in the exel. The handling for this problem is to add/fill

in the blank data so that the dataset is complete.

3. Transformation

In this transformation process, the data is transformed or combined into a format

that is suitable for processing in data mining. Transaction data will be made into

tabulation data in a boolean table presented in binary form, where the number 1 (one)

indicates the existence of a transaction and 0 (zero) indicates the absence of a transaction.

Application of FP Growth Algorithm

The FP-Growth algorithm is carried out using the RapidMiner Studio tool

application, the minimum support and confidence is determined by comparing support

and confidence values from the highest to the lowest so that effective support and

confidence values are obtained. The step to carry out the FP - Growth algorithm by

building the data structure used is Tree or commonly known as FP - Tree. The FP-Growth

algorithm is divided into three main stages, namely:

a. The conditional pattern base

b. FP-Tree conditional generation stage

c. Frequent itemset search stage

d. Results and Discussion

The results and discussions in the research are in the form of information generated

from the data mining process. The information generated is in the form of rules for

purchasing items that are often purchased by consumers at the same time, from this point

store owners can Siny.Co take a promotional strategy to find out product

recommendations for customers and store owners can find out which products can be

used as promotions for the future.

Clothing Product Recommendations Using the FP-Growth Algorithm in Siny.CO Stores

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3327

Results and Discussion

Data Selection

At this stage, the data used for the research will be selected to match the title of the

research to be researched. In this study, secondary data is used, namely transaction data

at Siny, Co stores with attributes consisting of dates, customer names, item names,

number of goods, and item prices.

Table 1

Clothing Store Transaction Data Before Pre-processing

Date

Customer Name

Item Name

Sum

Price

01/01/22

Ridwan

Tartan

2 pcs

IDR 200.000

02/01/22

Fitri

Sashi

1 pcs

IDR 125.000

03/01/22

Citra

Canna

1 pcs

IDR 125.000

04/01/22

Salman

Tartan

1 pcs

IDR 100.000

05/01/22

Fatimah

Alice

2 pcs

IDR 300.000

Table 1 shows the data of clothing sales transactions before the pre-pcessing stage.

Where the data is still intact when data is still being collected at the store Siny.Co each

number in the quantity attribute shows the number of clothes purchased.

Data Prepocessing/Cleaning

At this stage, the focus is on the cleaning process. The cleaning process includes

transaction data on Siny.co stores in 2022, removing data duplication and several

unnecessary attributes, including dates, customer names and prices. The removal of the

attribute was carried out because it was considered that it would not affect the rules of

association. Then, in this process, changes are made to the number of clothes purchased

replaced with 1 and the number of clothes not purchased is replaced with the number 0.

Tabel 2

Data Transaksi Toko Pakaian Setelah Pre – Processing

Transformation

At this stage, data transformation is carried out by providing initialization to

attributes that are adjusted to the type needed in the Fp-Growth algorithm. In this

transformation process, initialization is carried out on the attributes of the customer data,

which can be seen in Figure 2 below.

Customer

Tartan

Sashi

Alice

Canna

Duny Muhamad Firmansyah, Sunjana

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3328

Figure 2. Data Transformation

Application of FP–Growth Algorithm

In this study, the Fp-Growth algorithm is used in clothing sales transactions at

Siny.co stores shown in the figure below.

Figure 3. Application of the Fp-Growth Algorithm

Figure 3 above shows the processing of the sales transaction dataset using the

RapidMiner application. In this data processing using several operators, including:

1. Clothing sales transaction data retrive

The retrive operator is used to enter clothing sales transaction data after going

through the pre-processing process to be subsequently processed by the RapidMiner

application.

2. Numerical to binominal

This numerical operator serves to convert the type of numeric attribute to binomial.

Binominals are attributes that only have two values, namely true or false.

Clothing Product Recommendations Using the FP-Growth Algorithm in Siny.CO Stores

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3329

Figure 4. Numerical to binominal results

Figure 4 above shows the results of the numerical to binominal process, where the

attributes change to "true" and "false". A true value indicates a transaction while a false

value indicates no transaction.

FP-Growth

This Fp-Growth operator is used to identify the frequency of the itemset that will

be used by the operator later for the association process by determining the support value

of the sales transaction data at the Siny.co store. Below you can see the results of the

frequency itemset on clothing sales transactions with a support value of 20%.

Figure 5. Frekuent Itemset Results

Create Association Rule

This operator association rule is used to establish associative law by setting a

minimum confidence value of an item or itemset from a clothing sales transaction at a

Siny.co store.

Result

The results of the Association rule used in this study to determine the pattern of

clothing sales transactions in Siny.co stores using the RapidMiner application with a

minimum support of 20% and a minimum confidence of 80% resulted in 9 association

rules.

Duny Muhamad Firmansyah, Sunjana

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3330

Gambar 6. Graph Association Rules

In Figure 6 above is the result of the application of the Fp-Growth algorithm method

to clothing sales transactions at Siny.co stores in the form of graphs. Each rule is based

on the direction of the arrow.

Gambar 7. Description Association Rules

In Figure 7 above is the result of the application of the Fp-Growth algorithm method

to clothing sales transactions at Siny.co stores in the form of descriptions. Where there

are 9 rules that are eligible to build association rules on clothing sales transactions at

Siny.co stores and are used as the final rules in the dataset analysis. The 9 rules can be

explained as follows:

1. If you buy Canna products, then the possibility of buying Sashi products is also with

a confidence value of 0.879.

2. If you buy Alice products, then the possibility of buying Sashi products is also with a

confidence value of 0.911.

3. If you buy Tartan products, then the possibility of buying Sashi products is also with

a confidence value of 0.922.

4. If you buy Aruna products, then the possibility of buying Sashi products is also with a

confidence value of 0.949.

Clothing Product Recommendations Using the FP-Growth Algorithm in Siny.CO Stores

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3331

5. If you buy Cassandra products, then the possibility of buying Sashi products is also

with a confidence value of 0.952.

6. If you buy a Nala product, then the possibility of buying a Sashi product is also with a

confidence value of 0.959.

7. If you buy Lalita products, then the possibility of buying Sashi products is also with a

confidence value of 0.961.

8. If you buy Alice and Cassandra products, then the possibility of buying Sashi products

is also with a confidence value of 0.974.

9. If you buy Acio products, then the possibility of buying Sashi products is also with a

confidence value of 0.987.

Conclusion

Based on the research that has been carried out, a conclusion can be drawn from

this study, namely that the data of clothing sales transactions at the Siny.co Store can be

carried out association rules using the FP-Growth algorithm in the RapidMiner

application. From the application of the FP-Growth algorithm model, the sales transaction

data of the Siny.co Shop clothing was generated by 9 rules, including products that can

be recommendations for the Siny.Co Shop, namely Sashi, Canna, Alice, Tartan, Aruna,

Cassandra, Nala, Lalita, and Acio. Where these products can help the Siny.co Store to

recommend products that are of interest to customers so that they can launch a

promotional strategy for the Siny.co Store in the future.

Duny Muhamad Firmansyah, Sunjana

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3332

Bibliography

Allam, Zaheer, & Dhunny, Zaynah A. (2019). On big data, artificial intelligence and

smart cities. Cities, 89, 80–91.

Brous, Paul, Janssen, Marijn, & Herder, Paulien. (2020). The dual effects of the Internet

of Things (IoT): A systematic review of the benefits and risks of IoT adoption by

organizations. International Journal of Information Management, 51, 101952.

Hasan, Firman Noor, Aziz, Achmad Sufyan, & Nofendri, Yos. (2023). Utilization of Data

Mining on MSMEs using FP-Growth Algorithm for Menu Recommendations.

MATRIK: Jurnal Manajemen, Teknik Informatika Dan Rekayasa Komputer, 22(2),

261–270.

Hilman, Kirana Anastasya Afika Putri Hilman. (2022). Analisa Data Penjulan pada Toko

Kelontong Musyawarah Menggunakan Algoritma Apriori. Journal of Applied

Computer Science and Technology, 3(2), 221–227.

Kana, Fiqal, Ramadhan, Mukhlis, & Mahyuni, Rina. (2022). Implementasi Data Mining

Menganalisa Pola Penjualan Rempah-Rempah Menggunakan Metode Fp-Growth.

Jurnal Sistem Informasi Triguna Dharma (JURSI TGD), 1(4), 557–564.

Kumar, Mrinabh, & Dubey, Animesh Kumar. (2023). An analysis and literature review

of algorithms for frequent itemset mining. International Journal of Advanced

Computer Research, 13(62), 1.

Merliani, Nanda Nurisya, Khoerida, Nur Isnaeni, Widiawati, Neta Tri, Triana, Latifah

Adi, & Subarkah, Pungkas. (2022). Penerapan Algoritma Apriori Pada Transaksi

Penjualan Untuk Rekomendasi Menu Makanan Dan Minuman. J. Nas. Teknol. Dan

Sist. Inf, 8(1), 9–16.

Salu, Nerwan, Michael, Aryo, Padang, Samuel Yakobus, & Adda, Muhammad Sofwan.

(2022). Prediksi Persediaan Sepeda Motor Pada Dealer Yamaha Jaya Baru Motor

Mengunakan Metode Frequent Pattern (FF-Growth). INFINITY: UKI Toraja

Journal of Information Technology, 2(1).

Sari, Kumala, Muhammad Syahril, S. E., Kom, M., & Suharsil, S. E. (2021). Penrapan

Data Mining Untuk Pengelompokkan Data Member Restoran Dengan Menggunkan

Metode Hierarki Agglomerative Single Lingkage Pada Restoran Al-Jazeerah

Medan. Jurnal Cyber Tech, 4(2).

Saura, Jose Ramon. (2021). Using data sciences in digital marketing: Framework,

methods, and performance metrics. Journal of Innovation & Knowledge, 6(2), 92–

102.

Shawkat, Mai, Badawi, Mahmoud, El-ghamrawy, Sally, Arnous, Reham, & El-desoky,

Ali. (2022). An optimized FP-growth algorithm for discovery of association rules.

Clothing Product Recommendations Using the FP-Growth Algorithm in Siny.CO Stores

Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 8, August 2024 3333

The Journal of Supercomputing, 78(4), 5479–5506.

Wardani, Ni Wayan. (2020). Penerapan data mining dalam analytic CRM. Yayasan Kita

Menulis.

Wu, Yilan, & Zhang, Jing. (2023). Retraction Note: Building the electronic evidence

analysis model based on association rule mining and FP-growth algorithm.

Springer.