How can I effectively translate business problem into a Data Mining Task?



Please guide me with the things to focus on while translating the business problem into a data mining task like one of the business problem is “How to improve the performance of front line managers in an insurance company?”

Help would be highly appreciated?




Before working on any business problem, it is important to translate it to a data mining challenge. Look at the problem statement “How to improve the performance of front line managers?”, here it is difficult to measure what you want to achieve and the problem which is difficult to measure are hard to put a value on. Whenever possible, break the broad goals down into more specific one and link with monitoring value.

Before converting it to a data mining problem, we should take a structure approach to understand the detail about the business problem. Like in this case, what is productivity? Are you talking about total business sourced or profit? Let’s look at the below scenarios:

  • What do you mean by productivity, is it total business sourced in next 3 months (from month of hiring) or 6 months or 12 months.
  • What are factors is involved in productivity like product average premium, number of sales, product and other factor. Generate the equation first.
    Productivity = Number of product sales * Average premium of product
    Here productivity can be improved by increasing number of sales or by improving the ticket size.

Hope this helps!