Monday, November 1, 2010

Techniques for Data Analysis

So now that you have data, what do you do with them?  Data analysis, right?  But what is that anyway?  A useful way to think about data analysis is to relate it to driving a car. 

In a car, the rear view mirror tells the driver what is behind them.  The data equivalent for this is a historical report.  Typically historical reports display past trends and give a sense of how a metric of interest is changing.  The most easily interpreted analytic reports are those that display summary (i.e. rolled up) information that has an option for "drill down" into detail.  For example, a historical trend of summary sales results shows how they have changed over time, but yet allow for filtering and drill down to the detail behind the numbers.

You can't drive a car by looking only in the rear view mirror though!  Another useful tool to monitor your driving experience is the ever present dashboard.  This gives information about your current state such as speed, RMP, time, temperature, trip distance and fuel.  The analogy for this in the data world is its namesake, the data dashboard.  These data dashboards allow the business user to get a snapshot of the current state of affairs within a specific business function.  Like historical reports, the best dashboards display summary information with the option for filtering and drill down.  For example, a sales performance dashboard will summarize bookings, pipeline and backlog to date, with a comparison to a specific baseline or goal (think speed limit).  These dashboards are designed in a way that make for easy interpretation "at-a-glance", and are tailored for specific business roles.

A relatively new tool for drivers is the GPS.  This is the tool that allows the driver to specify their destination goal, such that they be provided step by step directions for reaching that destination.  The added benefit that the GPS provides is a fresh set of directions should the driver get off course or if conditions change.  A data analysis equivalent to the GPS are decision support tools.  These tools allow the business manager to specify their business goals (i.e. a sales target), and then indicate the specific decisions necessary to achieve these goals (i.e. sales staff to hire, prospects to contact, etc.).  Decision support tools are typically driven by data mining techniques such predictive analytics and optimization.

When the rear view mirror, dashboard and GPS fail, what do good (non-male) drivers do?  Stop and ask for directions, of course!  When a business user can not get the answers they need from reports, dashboards and decision support tools, ad hoc analysis can provide custom answers to these custom questions.  Ad hoc analysis by definition is customized to meet a specific need and as such requires specialized resources to extract, organize and analyze the data.  These analyses generally are the exception not the rule.  If they do become frequent, then consideration must be given to the usefulness of existing reports, dashboards and decision support tools.

While each of these tools have their use in supporting business operations, no one can claim dominance over the others.  Each are equally important (yes, just like their auto equivalents).  So get that car out of the garage and take your data for a spin!