An organisation has collected a large amount of data on various performance measures. Traditionally they have used tools in spreadsheets to produce simple summary statistics (eg mean, median), and to examine the effect of one factor at a time on the summary statistics of the performance measures. However, there can be several factors that can affect performance, and some of these factors are inter-related. So which are the most important factor(s) that affect performance.

Unlocking the Power of Data

We proposed several statistical techniques to look at the various data:

• Multiple regression to examine the relationship between several predictor variables and the performance. Models were built to try to explain these relationships, as well as how some of these variables can be used to predict future performance.
• Multivariate analyses to reduce the number of closely related variables into a smaller number of dimensions.
• Statistical Process Control (SPC) techniques to see the variations in the process that affect the overall performance, and to identify possible trends

Enhancing Knowledge and Performance

The analyses have produced some interesting results and provide a greater understanding on the variables that are more likely to affect performance. The results also contain a few surprises - certain variables that are previously considered as important (eg anecdotal comments, beliefs based on subjective views etc) actually turn out to be not significant in predicting performance.

The findings have generated further questions and hypotheses, and allow the organisation to look further into the data, as well as start more detailed research into the performance of some of the areas.