A Comparative Evaluation of Business Intelligence Technologies with Application to Product Profiling

Most of the Business Intelligence tools available on the market today have either been developed and industrially operationalised as “one size fits all” solutions or offered with multiple options leaving the business to decide on the best technology to use. We infer that this approach is likely to result in various analysis inaccuracies; hence rendering inappropriate business decisions. Accordingly, evaluating which technologies present more accurate results against a particular business need remains imperative. While using customer data from a large financial services company in South Africa, we analysed the performance of Neural Networks, Artificial Immune Systems and Bayesian Networks in classifying customer buying patterns. We measured the accuracy percentage values for a customer’s propensity to buy policies and also for existing policies lapsing. We observed that such assessments provide great insight in assessing the effectiveness of Business Intelligence enabling technologies. In particular, when applied to a larger data set, various customer patterns can be unearthed which results in adequate customer segmentation and business lead optimisation