LEE- Endüstri Mühendisliği-Yüksek Lisans
Bu koleksiyon için kalıcı URI
Gözat
Konu "customer segmentation" ile LEE- Endüstri Mühendisliği-Yüksek Lisans'a göz atma
Sayfa başına sonuç
Sıralama Seçenekleri
-
ÖgeRFMLP based customer segmentation and customer churn analysis in heavy equipment industry using customer transactions data(Graduate School, 2022-01-14) Çamlıca, Mustafa ; Çalışır, Fethi ; 507181139 ; Industrial EngineeringThe main target of this study is creating customer segments with customer transaction data of one of the leading heavy equipment industry companies, Borusan-CAT operating in Turkey which is a solution partner of Caterpillar Inc., and assigning churn probabilities to each customer. Customer transaction data is collected for the 2018 – 2020 period from complex database of the company. Data pre-processing step is completed in order to use raw data in this study. To decide the importance of the variables and customer segments Analytical Hierarchy Process was used 5 managers of the company respond a questionnare. After deciding the weight importance of the variables customer segmentation was completed with one of the unsupervised machine learning algorithms known as k-means clustering. 4 different customer segments were created. The importance of each customer segment was calculated with the help of weights that is result of Analytical Hierarchy Method. After customer segmentation, customer churn analysis was conducted. Churn analysis was completed with the help of supervised machine learning algorithms such as Logistic Regression, Support Vector Machines, Random Forest, and k-Nearest Neighbors. By comparing performance of each algorithm with the others, Random Forest was found as the most successful algorithm with highest accuracy rate in this study when it comes to predicting the customers who will churn in upcoming periods. There are no violations of the assumptions of each algorithm, therefore each of them can be used in this study. With customer churn analysis, each customer in the dataset labeled as churners or non-churners. Companies can use this information in order to complete such projects to prevent possible churners from churning in the future. Contributions of this study can be said as applying RFMLP based customer segmentation with a time-effective and efficient machine learning algorithm and applying customer churn analysis with the help of supervised machine learning algorithms to the customer transaction data of one of the biggest heavy equipment companies in Turkey. With this study 4 different customer segments are created and customer churn prediction is completed with high accuracy. Companies in the heavy equipment industry can utilize from this study to identify different customer groups and profile them, they manage their CRM and marketing strategies and allocation of resources can be completed with high effectiveness.