Use Cases

In retail banking, customer churn is a growing problem. One of our clients wanted to understand their own churn and prevent it from happening in the future. Using Rulex, they were able to get a holistic view of their customers and what drove them to leave. A variety of customer attributes were analyzed, and Rulex gave clear, conditional rules that explained how they related to churning. In the end, the bank reduced its churn rate by 26%.

Automotive companies, who often have a large and diversified product offering, need a way to manage their data on repair parts claims. TrucksCo wanted to predict which of their customers would need a repair part and when. Rulex integrated data about vehicles and past claims, and automatically extracted the rules that predicted when a vehicle would need a repair. Ultimately, TrucksCo improved both their customer assistance and their product.

Customers churn easily in the telecommunications industry. How can a Telco company retain its customers and remain competitive? Our Telco client wanted to predict when and why customers would churn. Rulex collected and merged data from the company’s website, call center, customer profiles, and usage plans. From this analysis, the company was able to understand which attributes correlated with a customer’s choice to leave.

Student retention is always a hot topic of discussion in colleges and universities. The decision to stay at an institution is influenced by many factors. Higher education professionals need to be able to identify student issues early and often, so they can know when and why a student might leave. Rulex analyzed data from 116,000 students over 9 years to predict which of their attributes (age, major, grades, and so on) correlated with dropout.

One of retail’s greatest challenges is accounting for shrinkage. The problem with conventional methods of controlling shrinkage, such as electronic surveillance and reporting systems is that they are reactive and rarely cost effective. Rulex helped a grocery retailer analyze a vast amount of seemingly unrelated data to determine the real causes of inventory loss. The retailer reduced its shrinkage significantly, by 43%.

Early prognosis is a central focus in cancer research. Healthcare providers need a reliable way to determine a patient’s prognosis in order to devise a treatment plan. In collaboration with one children’s hospital, Rulex was able to derive a set of four rules that identified a new class of poor-outcome patients who could benefit from new therapies.

One of an auto insurer’s main business goals is to reduce its amount of fraudulent claims. However, detecting fraud proves challenging when there is a massive data set of customer and claims history to be leveraged. Conventional machine learning seemed promising for our auto insurance client, but its high cost was a deterrent. Using Rulex, we assigned a set of attributes to over 200,000 auto accidents, from which we were able to increase fraud detection by 27%.

Monitoring networks in a cost-effective manner is a challenge facing all telecommunications providers today. As data volumes increase, companies need a reliable way to track bandwidth consumption. Our client, a very large Telco, faced this problem in an acute way. Using Rulex software, they gained insights they could apply to different aspects of their business. They increased bandwidth savings by 29%.

As gambling has grown over the years, so have behavioral disorders associated with problematic gambling. Identifying potential risk factors is crucial for planning preventive policies and therapeutic interventions. Rulex partnered with a university to identify those risk factors and predict their prevalence. Gamblers were interviewed, and information about their personal characteristics and type of gambling was collected. If-then statements about problematic gambling were derived from this data.

Hodgkin’s Lymphoma patients have a high likelihood of relapse despite recent advances in treatments. By combining clinical and genetic information, however, Rulex provided rules that helped physicians understand tumor biology and then address it with different therapeutic approaches. After using Rulex’s predictive model, the healthcare provider correctly classified 70% of its patients.