|
Abstract:
|
A successful statistical consulting process depended on a will trained, experience consultant with strong theoretical and applied knowledge of statistical models and tools in specific area. On the other hand it depends on a client with good research methodological background and basic statistical thinking capabilities. Practically, it is often hard to convey the gap between them. The proposed exploratory data analysis methodology can create a good corporation between the consultant and the client (researcher) that achieves more accurate results for the research. The approach makes better use of the available big data to define and refine the hypotheses and to build appropriate and more reproducible (repeatable) model for inference and prediction. The statistical consulting process can be divided to 4 stages: Stating and refining the research problem, exploring the data , building appropriate statistical model and interpret the results. The proposed approach will be applied in each stage, not only the stage of model building. So one compares the expected goal in each stage with the available information (data) and revise the expectation (or fixing the data) so the expectation and
|
