Diagnostic techniques

Any diagnostic technique’s main purpose is basically to discover the root cause of the problem or event you have discovered.

There are a few methods that can be employed in the oil price vs. profit scenario to help identify what the root cause is:

• Visualisation through trend and outlier analysis via:

  • Time series charting

  • Scatter plot

  • Heat map

• Binning techniques to help calculate and use cohorts to establish baseline targets for measuring against each B2B customer

• Applying target or tolerance levels to help discover outliers

• Shaping data for analysis – If the diagnostic analysis is not quite providing clear guidance on the root cause then perhaps additional data is required to blend into the existing dataset?

Trend analysis helped discover issues with the profit expectation but it can also help with the diagnostic analysis. By displaying each B2B customer trend versus the overall average profit margin expressed as a percentage difference, you could then rank the customers from lowest to highest margin and investigate the lowest margin ones further. Tableau can easily accomplish this with table calculations and LOD expressions.

Outlier analysis can help discover anomalous data points at a certain point in time or within an overall range of data. A scatter plot based on profit margin by total supply (bbl) per B2B customer helps discover any outliers from a cohort perspective. For example, high supply customers would naturally have a higher $ profit margin than the smaller supply customers but if a higher supply customer shows its profit margin is below the average or target profit margin for its cohort it could be considered an outlier.

If you prefer, all of these techniques can be complimented by an Ishikawa diagram which can help with the documentation of cause-and-effect of the issue being diagnosed.