Vis enkel innførsel

dc.contributor.authorLimseth, Amanda Cecilie Limseth
dc.date.accessioned2020-11-05T08:23:31Z
dc.date.available2020-11-05T08:23:31Z
dc.date.issued2020
dc.identifier.urihttps://hdl.handle.net/11250/2686489
dc.descriptionMasteroppgave(MSc) in Master of Business Analytics - Handelshøyskolen BI, 2020en_US
dc.description.abstractBusinesses will always have a certain amount of customers leaving for various reasons, whether it's for another competitor or because their perception of value isn’t met. The concept of churn prediction is based on investigating the pattern of behavior that has shown to lead to customers not renewing their membership in the past, and by identifying similar behavior in the current customer base to find the at-risk customers before they churn. There exists large variability in the customer usage under contract, both in the total number of visits and in the proportional type of activity attended. The study will use two different statistical approaches to construct a binary model for how usage under contract is linked to churn. Both approaches will differentiate between the usage amount and type of activity under contract to ultimately explain and predict the likelihood of churn in the fitness industry. The results from the binary prediction model indicate that there is a significant difference in the estimated likelihood of churn in the fitness industry based on the customer usage. It found that members with monthly averages over the contract period at the mean or higher is generally less likely to churn and less affected by the change in explanatory variables like activity type, age, and proportional usage in the last months leading up to the renewal decision. The members that showed lower usage over the contract period had churn likelihoods that were considerably more affected by the change in usage variables. The research will provide a statistical model that can be altered in order to work for any business that has contractual memberships where the usage amount and type of activity is recorded. The model will provide churn prediction probabilities before they occur, which can enable businesses to take targeted action ahead of time and attempt to lower their associated customer attrition by creating tailored promotions to entice the customer to renew their contracts.en_US
dc.language.isoengen_US
dc.publisherHandelshøyskolen BIen_US
dc.subjectbusiness analyticsen_US
dc.titleChurn Prediction: How Customers' Usage under Contract is Linked to Churnen_US
dc.typeMaster thesisen_US


Tilhørende fil(er)

Thumbnail
Thumbnail

Denne innførselen finnes i følgende samling(er)

Vis enkel innførsel