but before you do, make sure youve defined your role. if you work for a large firm, youll be answerable to managers and middle managers who operate within the firm. they dont necessarily understand the concepts you work with or the problems you try to solve. they also dont have the resources necessary to drive your efforts to fruition.
in fact, if youre fortunate enough to work at a startup, you might be responsible for everything. these companies usually are not structured so much as developing and deploying software, and the people with the most direct contact with the user community are the project managers, who often have accounting and finance backgrounds. they often have neither the time nor the necessary training to do anything more than write software. this is also the case in the increasingly popular open source software movement. both these types of organizations have little, if any, place for a business analyst.
while traditionally statistical modeling is done to optimize an existing product or process, you must understand the limitations and benefits of statistical modeling. if you are doing statistics to maximize the quality and yield of a product or process, are you simply subjecting workers to too many hours of production? or is statistics the only possible solution? in the latter case, is the statistical model functioning as an independent optimization or as a predictive tool? in the former, is the process producing the optimized quality or is it merely a by-product? if statistics is used to make decisions, are those decisions made based on cost or quality? if statistics is used to predict quality, is a continuous or binomial distribution appropriate? if statistics is used to optimize quality, is a linear regression the best approach or is an exponential model? you must research the approaches before embarking on a project. many models are available to guide your decision process.