TY - GEN
T1 - Retention analysis based on a logistic regression model
T2 - 15th IEEE International Conference on Networking, Sensing and Control, ICNSC 2018
AU - Ghahramani, Mohammadhossein
AU - Zhou, Mengchu
AU - Hon, Chi Tin
AU - Wang, Gang
N1 - Publisher Copyright:
© 2018 IEEE.
PY - 2018/5/18
Y1 - 2018/5/18
N2 - Telecommunication data has provided new opportunities for both businesses and academia to analyze subscribers' behavioral patterns. Recently, there have been many changes in this industry, i.e., lessening of market regulations/restrictions in exchange for greater participation. New services, emerging technologies, and competitive offerings are factors causing customers to move to different companies. In this work, we intend to develop a logistic regression model tailored for a telecommunication company in Macau by forecasting potential subscribers intending to leave their current services. To implement such prediction we should assign a probability value to subscribers, based on a relationship between customers' historical data and their future behavioral pattern. Then customers with the highest propensity to leave can receive various marketing offers. To improve the analysis result we have utilized a combination of two datasets. Our experimental results show how such data aggregation can improve the model accuracy.
AB - Telecommunication data has provided new opportunities for both businesses and academia to analyze subscribers' behavioral patterns. Recently, there have been many changes in this industry, i.e., lessening of market regulations/restrictions in exchange for greater participation. New services, emerging technologies, and competitive offerings are factors causing customers to move to different companies. In this work, we intend to develop a logistic regression model tailored for a telecommunication company in Macau by forecasting potential subscribers intending to leave their current services. To implement such prediction we should assign a probability value to subscribers, based on a relationship between customers' historical data and their future behavioral pattern. Then customers with the highest propensity to leave can receive various marketing offers. To improve the analysis result we have utilized a combination of two datasets. Our experimental results show how such data aggregation can improve the model accuracy.
UR - http://www.scopus.com/inward/record.url?scp=85048234234&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85048234234&partnerID=8YFLogxK
U2 - 10.1109/ICNSC.2018.8361375
DO - 10.1109/ICNSC.2018.8361375
M3 - Conference contribution
AN - SCOPUS:85048234234
T3 - ICNSC 2018 - 15th IEEE International Conference on Networking, Sensing and Control
SP - 1
EP - 6
BT - ICNSC 2018 - 15th IEEE International Conference on Networking, Sensing and Control
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 27 March 2018 through 29 March 2018
ER -