Business Analytics Applications in Budget Modelling To Improve Network Performance Eleni Rozaki Institute of Technology Tallaght, Dublin Ireland
Key Principles of the Project Integrate all Network Data Sources Mobile Network Performance Management Cost Considerations and Customer Segmentation End to End Optimisation using Business Analytics Budget Planning Classification Rules using Predictive Analytics
Data Sources in Mobile Networks Rate of Revenue TECHNICAL Customer churn Revenue Cost Texts Calls Networking Optimise Manage Customer Profiles CRM Billing Predict Business Analytics Rate of cost of optimisation services
Automate the Network Optimisation process
Cost Considerations and Network Performance Estimate network optimisation costs , overheads and estimated revenue Build a data mining model and estimate customer willingness to pay For each customers deal, estimate the number of users Using the estimates, determine the profit for mobile services Choose the most profitable deals, prioritise customer profiles based on their needs
End to End Optimisation Using Business Analytics
End to End Optimisation Plan to Track Expenditures Dynamic profiling & enhanced customer segmentation Personalization of customer deals based on their needs Identify newly emerging customer issues Customer churn prediction Budget planning-track expenditures
Review an efficient budget process and improve expenditure prioritisation Network Data Budget planning-Track expenditures Reporting / Analytics Revenue Data Enterprise BI and reporting Applications Identify future revenue and expenditure trends Call Details Enterprise Analytics Applications Web and Mobile Text Data Customer ERP/CRM Data Customer web And mobile Apps App Data Mobile App Server Internet Data Accuracy of Forecasts Channel Campaigns Optimisation Plan Marketing Campaign Mgt Mail Costs SMS Network Optimisation Cost Print POS Marketing Cost Budget Planning
Clustering Techniques Research Methodology - Data Mining Models Cost Sensitive Classifiers Neural Networks Clustering Techniques Deals Customer profiles Location Demographics Offers Costs
An End to End Optimisation Model Using Clustering Techniques Dynamic profiling Personalisation of customer deals Identify newly customer issues
A Budget Model Using Neural Networks Budget planning-forecasting Identify future revenue and expenditure trends Cost efficiency in mobile networks Financial and operational planning
Data Mining Models Cost Sensitive Classification Clustering Neural Networks Classification Clustering Association Rules Relations & Patterns Prediction Unsupervised Model Sequence Discovery Backpropagation algorithm
Costs in Machine Learning The learning algorithms of our cost model: Minimising classification errors Different types of misclassification have different costs based on the customer deals Provide more informative output of performance assessment under highly imbalanced data Make the prediction that minimises the expected cost
Results Clustering Neural Networks Pros: Summarising information about customers profiles and customers deals. 77.5% accuracy Cons: Numerical variables Neural Networks Pros: Hidden relations between the different mobile deals and users profiles. 87.5% accuracy Cons: Risk of overfitting the model
Conclusions 1 The strategic use of business analytics could bring major benefits to mobile service providers 2 Mobile providers need network optimisation solutions that are embedded in business analytics to correlate information from different sources 3 Data such as services, device portfolios, cost and billing and network service quality can be used for efficient customer segmentation 4 Business analytics helps with financial planning and ensures that network optimisation costs and budgets are accurate