2022
DOI: 10.1109/tg.2021.3067114
|View full text |Cite
|
Sign up to set email alerts
|

RFM-LIR Feature Framework for Churn Prediction in the Mobile Games Market

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
2
1

Citation Types

0
5
0

Year Published

2022
2022
2025
2025

Publication Types

Select...
5
2
1
1

Relationship

0
9

Authors

Journals

citations
Cited by 18 publications
(5 citation statements)
references
References 38 publications
0
5
0
Order By: Relevance
“…From the perspective of prediction methods, these are mainly divided into traditional machine learning and deep learning categories. In the application of traditional machine learning methods, Perisic and others have used logistic regression to predict mobile game users [2], Ullah and others have utilized random forests for telecom user predictions [3], Wang and others have applied Gradient Boosting Decision Trees (GBDT) for predicting user churn in advertising platforms [4], Swetha has employed XGBoost for telecom user predictions [5], and Khan and others have adopted artificial neural networks for telecom user churn predictions [6]. Additionally, researchers have studied the performance of different machine learning algorithms on the same dataset [7][8][9][10][11][12][13][14].…”
Section: Related Workmentioning
confidence: 99%
“…From the perspective of prediction methods, these are mainly divided into traditional machine learning and deep learning categories. In the application of traditional machine learning methods, Perisic and others have used logistic regression to predict mobile game users [2], Ullah and others have utilized random forests for telecom user predictions [3], Wang and others have applied Gradient Boosting Decision Trees (GBDT) for predicting user churn in advertising platforms [4], Swetha has employed XGBoost for telecom user predictions [5], and Khan and others have adopted artificial neural networks for telecom user churn predictions [6]. Additionally, researchers have studied the performance of different machine learning algorithms on the same dataset [7][8][9][10][11][12][13][14].…”
Section: Related Workmentioning
confidence: 99%
“…As shown by Alves Gomes et al [36] most customer representations are modeled with manual features extracted by experts or with the RFM analysis [37]. For example, Perisic et al [38] and Friedrich et al [39] extracted RFM-based features by extending the RFM analysis from historical data for customer representation. Wu et al [40] modeled and analyzed customer behavior with an extended RFM approach by adding customer contribution time and repeat purchase attributes and combining it with a k-means clustering.…”
Section: Customer Representationmentioning
confidence: 99%
“…Marín et al modeled user behavior based on the traditional proximity, frequency and currency (RFM) model to obtain a proximity, frequency, importance and duration (RFID) model of customer assessment from the perspective of customer-contact center interactions, and showed that the model can be generalized to any environment requiring classification or regression algorithms in any environment that requires classification or regression algorithms [15]. Perišić et al proposed an extended framework of new proximity, frequency, and monetary value features for predicting user churn in the mobile gaming domain by combining features related to user lifecycle, intensity, and rewards, and indicated that the top five most important features of a multivariate churn prediction model include long-term and short-term frequency features, monetary, intensity, and lifecycle features [16]. Wei et al first established an RFLP metric system for predicting MOOC user learning behavior and attrition by improving the RFM model in the business domain; secondly, histogram and chi-square tests were used to determine the characteristic variables affecting MOOC user attrition; finally, a MOOC user attrition prediction model was constructed by combining the Grouped Data Processing (GMDH) network as a post-processing information system [17].…”
Section: Related Workmentioning
confidence: 99%