|
Hyperparameters Tuning and Model Comparison for Telecommunication Customer Churn Predictive Models
|
|
| Author:
|
KOH GUAN LI, BOOMA POOLAN MARIKANNAN
|
| Abstract:
|
Customer churn is a phenomenon that occurs when a customer cases doing business with a company or service.
Thus, predicting customer churn is one of the primary aim of businesses in various sectors especially those which
involve subscription-based services such as telecommunication sector. The markets of telecommunication sector
are also very saturated which led to high competitiveness among competitors. Various researches had been
conducted to develop churn predictive models utilizing various algorithms to accurately predict churn in the
telecommunication sector. However, not much emphasis has been allocated in tuning the hyperparameters of the
developed algorithms or models in order to further improve the performances of churn predictive models. This
paper proposes the development of churn predictive models using the Naive Bayes (NB), Decision Tree (DT) as
well as Artificial Neural Network (ANN) algorithms along with the incorporation of hyperparameters tuning
through grid search to look for the best hyperparameters' setting which can optimize the performance of each
algorithm. Grid search method manually searches a specified subset of the hyperparameter space of a machine
learning algorithm and identify the best setting of hyperparameters within the designated subset which can improve
the overall performance of the algorithm or model.
|
| Keyword:
|
Naive Bayes, Decision Tree, Artificial Neural Netowork, grid search, customer churn prediction
|
| EOI:
|
-
|
| DOI:
|
-
|
| Download:
|
Request For Article
|
|
|