user assistance in data mining

There are a few stumbling stones preventing users and data analyst from reaching their goal of knowledge discovery in databases.

1. Task identification, which is often thought to be clear. This could also involve business understanding and data understanding on an iterative basis.

2. Selection of proper algorithm, model

3. Evaluation criteria (understandability, accuracy, algorithm complexity etc) and tradeoffs between: accuracy/speed/MDL (minimum description length)/cost of false positive/negative and so on.

Recent work of mine will be proposing a framework to facilitate user to manage their expectation and control deliverable result from mining effort.
to be continued