Akademik Veri Yönetim Sistemi
3rd International Conference on Data, Electronics and Computing, ICDEC 2024, Kayseri, Türkiye, 18 - 20 Eylül 2024, cilt.1455 LNNS, ss.433-447, (Tam Metin Bildiri)
In this study, customer complaints from three different competing furniture companies were analyzed using text mining methods with Python programming. The aim of the study is to process unstructured linguistic data using natural language processing (NLP) techniques to correlate complaints with call types. In this context, a total of 9,930 complaints obtained from three different furniture companies between 2022 and 2023 underwent preprocessing, NLP, topic modeling, and classification processes. The data processed in preprocessing and NLP steps were grouped into four main topics using Non-Negative Matrix Factorization (NMF), an effective method for extracting meaningful topics from large and complex datasets. The results provided important insights into areas where furniture companies experience customer dissatisfaction and the specific issues that complaints focus on. Identified topics were evaluated for their alignment with call types identified by customer representatives using topic modeling techniques. Finally, the identified topics were designated as target variables for each complaint and classified using various classification algorithms to create a call assignment model. When comparing the performance of classification algorithms in this model, Support Vector Machine (SVM) was identified as the algorithm with the highest accuracy.