Company: IBM BAO
Authors: Sanket Jain
Date: 29/11/11
The growing number of data being generated at Telecom companies dues to their customers doing interactions with their friends/followers where they use their mobile phones to make voice calls/SMS has caused the data in Telecommunications industry to increase at a very rapid pace. There is also a growing interest in integrating the Telco customer base with social media platforms like Facebook. Given that there are already 700 million customers on Facebook,the task of correctly tying the customer mobile phone number to his Facebook account in itself will be a tedious one. Due to all these factors, the data will become difficult to manage if a Telco service provider wants to do advanced analytics to do churn predictions, build customer insights,etc. Also, an analyst would get tempted to use sampling for all his data mining tasks so as to reduce the time and effort for doing his analytical tasks without the loss of generality, and he might also be tempted to use as small a sample size as possible. This can result in a large value of the standard error. Next, whether or not you can apply sampling to areas like churn analysis has been discussed. This study also does a literature review of some of the sampling techniques being followed in Telecommunications industry and in some of the other industries. In the end, it has been suggested how sampling can help telecom service providers to use credit bureau data for building cross-sell/up-sell propensity models, risk models, fraud detection models, etc. This paper is dedicated to TM Forum. This work will be of interest to those considering and planning how to use social networking analysis and social marketing analysis in Telecom domain to
augment their customer analytics activities concerned with customer care and churn management. |