Business intelligence and analytics

Analyitics and business intelligence is most often connected to a gap between the current situation and a goal. To start analysing this problem, we select one or several data variables of interest.

With sql or data integration tools we retrieve data for the population and variables we want to look into.

Visualize
The next step is to visualize the variables of interest. For some data sets plotting a variable over time gives insight. For instance we can put sales on the Y axis and month on the X axis. In other visualizations we put the dependent variable on Y and independent on X to detect patterns. Cross tables can give an indication of correlation and causation. Correlation coefficients will detect linear relations between variables. We will also look at average, standard deviations, max, and min.

Open source application for bayesian nettwork with data from a Coursera Big Data and Web Analytics course.

Think
Based on the initial discovery of how the most important variables behave, we will think trough what additional variables could be causes and what kind of tools that could identify those. Simple regression could be one method. Path analysis with chains of several variables, decision trees, bayesian networks, clustering, or causal loop diagrams are other possibilities. At this point we also have to reflect further about the problem or challenge, how valuable the insight from futher analysis is, and in what decisions it will be used. The initial findings could be all we need to make some experiments with variables we assume are determining factors. Such experiments will produce new data that we can analyse further.

OLAP-table (Online analytical processing) where data for Google Preview (snapshots) is retrieved.