Government officials wanted to reduce wait times for their community but needed a data scientist to lead the way.Data Science
Humberto leads a large Government department, tasked with overseeing the distribution and regulation of social benefit payments to its citizens. The department has more than 65 years of uninterrupted activity, it employs a big multidisciplinary team of professionals and lately has engaged into modernising its activities.
The current process has been causing financial and social hardship to those lodging an appeal. It wasn't uncommon for it to take 4 months until a decision was made.
The team figured out that by identifying and predicting likely claimants early, it was possible to substantially decrease their response times, thus providing a much better service.
Humberto engaged Jose to explore the transaction databases and apply different machine learning models.
After trialling different models, Jose proposed a system of classification trees that were able to predict with a success rate of 85% who will put in a claim.
The first thing Jose did was explore the existing data to get a solid handle on the underlying factors that were influencing the delays.
Ultimately it was decided that classification trees were the most appropriate solution for the problem. This method allowed the model to be communicated and understood without becoming over complicated.
The proposed model can predict with a success rate of 85% who will put in a claim, which means a significant reduction in waiting time and financial hardship for 85% of the people lodging an appeal.
Jose's data model has provided a better idea of the actual time that each case requires to be solved and as such the department can plan the staffing requirements more accurately and efficiently. The results of this pilot were deemed of such good quality that the implementation of them into the production services in underway.