Klasifikasi Churn Pelanggan Pada Perusahaan Industri Logistik Dengan Metode Decision Tree, Logistic Regression Dan Logit Leaf Model

Salsabila, Renada Aulia (2021) Klasifikasi Churn Pelanggan Pada Perusahaan Industri Logistik Dengan Metode Decision Tree, Logistic Regression Dan Logit Leaf Model. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of RDIB_TA_05211740000061_RENADA AULIA SALSABILA.pdf] Text
RDIB_TA_05211740000061_RENADA AULIA SALSABILA.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (1MB) | Request a copy

Abstract

Pelanggan merupakan salah satu parameter utama bagaimana bisnis dapat diinisiasi dan diukur perkembangannya. Dalam menjaga hubungan dengan pelanggan telah dijelaskan pada ilmu CRM(Customer Relationship Management), bahwa menjaga relasi antara perusahaan dan pelanggan akan mempertahankan loyalitas pelanggan sehingga dapat membuat keberlangsungan dan keberlanjutan bisnis tetap terkendali. PT.XYZ merupakan salah satu badan usaha milik negara yang bergerak dibidang logistik. Perusahaan tersebut terbilang telah memiliki cukup banyak layanan jasa dalam mendukung bisnis utamanya sehingga mampu menarik berbagai macam pelanggan. Tetapi pada kenyataannya berdasarkan KPI(Key Performance Indicator) atau target perusahaan saat ini, jumlah keuntungan dan total pendapatan yang diterima perusahaan tidak berbanding lurus dengan banyaknya jumlah pelanggan. Berdasarkan data laporan transaksi pelanggan real-time, diambil hipotesis bahwa beberapa pelanggan teridentifikasi bersifat idle, yaitu keadaan dimana pelanggan hanya diam dan berperan sebagai tamu yang tidak melakukan proses transaksi hingga pembayaran terjadi atau selesai dalam jangka waktu tertentu. Selain itu, istilah churn pelanggan atau bisa disebut sebagai pelanggan cabutan merupakan kondisi atau tingkat presentase pelanggan yang berhenti melakukan transaksi diukur dari kapan terakhir kali transaksi hingga periode waktu yang telah ditentukan. Ditambah berhubung juga dengan kondisi perusahaan yang sedang dalam tahap transformasi digital, maka dari itu perlu dilakukan analisis lebih mendalam terkait perilaku pelanggan pada saat transaksi, salah satunya adalah mengantisipasinya dengan memprediksi kecenderungan churn pelanggan. Penelitian dilakukan dengan data transaksi pelanggan dalam rentang waktu periode dari April 2020 hingga Desember 2020. Pra pemrosesan data diolah berdasarkan data deret waktu transaksi bulanan dengan penentuan variabel prediktor yang mencerminkan perilaku transaksi pelanggan, khususnya LRFM. Metode yang digunakan adalah teknik pembelajaran mesin, yaitu Decision Tree, Logistic Regression dan model hibrida atau kombinasi antara kedua metode terkait, Logit Leaf Model. Berdasarkan percobaan beberapa teknik pengulangan sampel data didapat hasil akhir berupa model dengan pengukuran kinerja menggunakan akurasi sebesar 88%, dan grafik ROC(Receiver Operating Characteristics) beserta nilai AUC(Area Under Curve) sebesar 0,89 yaitu Decision Tree dengan teknik pengulangan sampel data SMOTE-ENN.
====================================================================================================
Customers are one of the main parameters of how a business can be initiated and its development measured. In maintaining relationships with customers, it has been explained in the science of CRM (Customer Relationship Management), that maintaining relationships between companies and customers will maintain customer loyalty so that they can keep business continuity and sustainability under control. PT. XYZ is one of the state-owned enterprises engaged in in logistics. The company has quite a lot of services to support its main business so that it is able to attract various kinds of customers. But the reality shows based on KPI (Key Performance Indicator) or the company's current target, the amount of profit and total revenue received by the company is not directly proportional to the number of customers. Based on real-time customer transaction report data, it is hypothesized that some identified customers are idle, a situation where customers are silent and act as guests who do not process transactions until payment occurs or is completed within a certain time. In addition, the term customer churn or can be referred to as withdrawn customers is a condition or percentage level of customers who stop making transactions, measured from the last transaction until a predetermined period. In addition, considering the situational inside which is in the digital transformation stage, it is necessary to conduct a more in-depth analysis of customer behavior during transactions, one of which is to anticipate it by predicting the tendency of customer churn. The research was conducted with customer transaction data in the period from April 2020 to December 2020. Pre-processing data is processed based on monthly transaction time series data by determining predictor variables that reflect customer transaction behavior, especially LRFM. The method used is one of machine learning techniques namely Decision Tree, Logistic Regression and a hybrid model or a combination of the two related methods, Logit Leaf Model. Based on the experiment of several data sample repetition techniques, the results shows that the best model gain with performance measurement using an accuracy of 88%, and a ROC (Receiver Operating Characteristics) graph along with an AUC (Area Under Curve) value of 0.89, namely Decision Tree with the SMOTE-ENN technique.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Churn Prediction, Machine Learning, Decision Tree, Logistics Industry
Subjects: Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression
T Technology > T Technology (General) > T57.5 Data Processing
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User: Renada Aulia Salsabila
Date Deposited: 08 Sep 2021 07:29
Last Modified: 08 Sep 2021 07:29
URI: http://repository.its.ac.id/id/eprint/91811

Actions (login required)

View Item View Item