The proper analysis for data begins with an understanding of how the data is, or will be, obtained. Often, there are some restrictions to the randomization of the experiments, or the data is observational. This dictates the method of analysis. Using our broad background in applications, we construct the appropriate model. From the model, the method of analysis is determined. In the case of design of experiments, this permits a re-design of the experiment, if necessary, to facilitate answering the important questions. The statistician works closely with the client to determine and formulate these questions. The statistician then uses the appropriate statistical software, and explains the conclusion using the client’s terminology.
In the case of very large data sets, Data Mining gives several options. One may develop a predictive model for either binomial or continuous outcomes. Also, one may develop a classification scheme, perhaps, using cluster analysis or a decision tree.
