WHAT WE DO
At OLSPS Analytics, we pride ourselves on being able to derive predictive analytic solutions across a broad spectrum of problems in a wide variety of industries. Our highly skilled team of data scientists has the ability to develop a customised solution to fit your particular problem. Examples of solutions that we have recently developed include:
- Retailer Credit Risk
- Medical Aid Fraud
- Market Basket Analysis
- Next Best Offer
- Claims Segmentation
- Supplier Segmentation
- Churn Modelling
- Credit Scorecards
Churn models are built in order to predict the likelihood of a change in customer behaviour in a given timeframe. Users of churn models are primarily concerned with predicting the likelihood of a current client ‘churning’ (discontinuing the use of their service). Churn models are also able to calculate the estimated financial loss to companies as a result of the potential client churn. The output of a churn model will allow the user to design a targeted retention campaign which is aimed at preventing high value, high risk clients from churning.
Segmentation is a versatile and powerful tool which enables users to split a large group (of companies or people or similar) into smaller groups which exhibit similar behaviours and patterns. As a result, interactions with those smaller groups can be customised and adapted by the segmentation user so that they are specific to each identified category. This allows the segmentation user to make the interaction more personalised with fairly limited effort.
Another type of segmentation solution segments suppliers based on their relative cost, location, BEE status and other pertinent factors. Using this solution, companies are able to assess the performance of a supplier based on a number of objective measures, allowing them to make consistent, justifiable choices to allocate their business amongst their suppliers. The solution has been developed based on allocating insurance claims to a specific panel beater, but can be easily extended to many different industries.
Scorecards enable users to rank individuals relative to a target. The scorecard is built by evaluating each individual’s score on specific variables that are related to the target. The most common application of scorecards is to assess customer credit risk using credit history and other demographic variables as the inputs.