In today’s campaigns management, a set of target customers are selected according to pre-defined scrores derived by data mining or business analytics techniques. While in principle the vast and ever-growing amounts of available historical data would allow for deriving increasingly precise models, the complexity resulting from the consideration of large volumes of multivariate, fine-grained data leads to the fact that dependencies and relationships within the data are not found, algorithms do not scale, and traditional statistics as well as data-mining techniques collapse because of the well-known curse of dimensionality. Nowadays, these problems are often referred to as big data symptoms. In addition, in practice various business rules and guidelines are to be considered, in particular regarding the frequency and channel an individual customers should be contacted; currently, such types of constraints lead to combinatorial problem not supported by business analytics tools.
This projects targets at triggering specific campaigns for individual customer based on observed customer dynamics and rules learned from empirical data, considering constraints business guidelines. In collaboration with a large European telecommunications company, we are looking at
- - Intelligent (in terms of campaign specific) data dimensionality reduction and feature extraction
- - Temporal and sequentual pattern mining
- - Principled approaches to customer segmentation
- - Customer dynamics and cluster-transition modelling
- - Temporal, combinatorial rules for campaign initiation
- - Density and angle-based anomaly and novelty pattern detection