Performance Comparison of Classifiers for Prospective Buyers Identification in ethio telecom Mobile Cross-Selling Market
| dc.contributor.advisor | Epherm, Teshale (PhD) | |
| dc.contributor.author | Getachew, Gemechu | |
| dc.date.accessioned | 2020-03-09T06:13:07Z | |
| dc.date.accessioned | 2023-11-04T15:13:12Z | |
| dc.date.available | 2020-03-09T06:13:07Z | |
| dc.date.available | 2023-11-04T15:13:12Z | |
| dc.date.issued | 2020-02-22 | |
| dc.description.abstract | Direct marketing is a form of communicating an offer directly to a targeted group of customers through a variety of media. It plays a major role in customer retention and service provisioning tasks. Retaining customers by providing products and services that meet their need is one of the main objectives of customer relationship management. Identifying prospective customers for direct marketing enables a company to reach specific audiences which will more effectively respond to promotions. Moreover, direct marketing helps businesses to optimize their marketing budget, keeps current customers loyal to them, and makes businesses capable of measuring the result obtained from promotions. Ethio telecom promotes service packages to its customers through SMS and mass communication channels. However, promotions should target customers based on the specific services they use, and customer over-touching should be reduced especially during SMS advertisement. In the current practice, no scientific methodology is implemented to estimate the potential respondents to cross-selling market promotion. Promotions are communicated to both potential buyers and non-buyers without distinguishing the two groups. Direct marketing approaches help the company to effectively allocate resources and give services based on the interests of customers. The aim of this thesis is to identify prospective customers in ethio telecom mobile valueadded service market. To achieve this goal, five classifiers namely Naive Bayes, Neural network, SVM, K-nearest neighbour, and Decision tree (J48) tested with customers service usage historical data. In this process, 900,000 customers’ actual CDRs from ethio telecom were gathered and raw data aggregated with the aim of representing users’ behaviour. The representation was based on users’ responses towards service fee and time preference to use services. Sixteen feature variables and one predictor variable are constructed from the raw CDR collected. Data cleaning and class balancing done, and the selected classifiers tested for their accuracy in identifying prospective buyers of service packages. The right customers for direct marketing are identified and ways to minimize customer over-touching during the promotion of low-price packages are shown. So, beyond selecting a classifier with high accuracy, the study aims at maximizing correctly classified instances. Accordingly, except Naïve Bayes classifier, the other four classifiers resulted in better performance. with Neural network classifier more than 90% of voice and 93% of SMS packages potential buyers are identified. whereas Decision tree (J48) has scored the best result with data package buyers dataset by identifying more than 94% of potential buyers. | en_US |
| dc.identifier.uri | http://etd.aau.edu.et/handle/123456789/21050 | |
| dc.language.iso | en_US | en_US |
| dc.publisher | Addis Ababa University | en_US |
| dc.subject | Cross-selling | en_US |
| dc.subject | Classification techniques | en_US |
| dc.subject | SVM | en_US |
| dc.subject | Neural Network | en_US |
| dc.subject | KNN | en_US |
| dc.subject | Decision Tree(J48) | en_US |
| dc.subject | Mobile value-added service market | en_US |
| dc.title | Performance Comparison of Classifiers for Prospective Buyers Identification in ethio telecom Mobile Cross-Selling Market | en_US |
| dc.type | Thesis | en_US |