Settings

See below details about the Settings menu in the left menu of DataMa interface:

  • Level of aggregation (NumLimit): The level of aggregation that the model is using – e.g. if Level of aggregation is set at X%,  segment within each dimension that represents less than X% of the Primary Numerator (e.g. Revenues) of the main KPI you’re analyzing will be clustered in one « Other » segment. X is set at 2 by default, but you may want to play with this parameter quite a bit because it can change significantly the calculation of mix effects.

 

  • Smart Dimension: instead of having a Dimension analysis made independently on each Dimension or Combining all dimensions together, Smart Dimension assesses all pairs of Dimension to raise the pairs that explain the much the variation you are trying to understand.

 

  • Skipped steps (selected): the steps that are kept within your overall Metric Relation definition. This allows you to change your overall « Market Equation » and the KPI you’re following, as well as grouping/ expanding the steps you’ve defined

 

  • Split Mix and Perf: By default DataMa split each step by Mix & Perf but you can disable this split. If so DataMa will only make a split by Positive and Negative Segments.

 

  • Impact tests: if you have DataMa IMPACT license you can directly test the significance of a variation within DataMa COMPARE. You will just have the result of the test, the detail will be accessible only via the IMPACT interface.

 

  • Safe Mode (SafeMode): Activates two important checks:
    • The volume of events you’re considering in your analysis (DimensionForFlag, MinValueForFlag): Is the total sum of selected Dimension above a certain threshold both within Start and End ? The idea is to check that your analysis is « significant » and everything you’re saying makes some sense. If not, a flag will be raised. Of course, this is far from a proper statistical significance test. DataMa Impact might help you in doing that properly, with the appropriate statistical tests
    • The correlation between your dimension is not too high (DimensionsAreNotIndependent). We’re using Chi-Square test here to evaluate the correlations within dimensions. This is important because when you compute a mix effect on two dimensions, it could very well appear that those two mix are actually the same effects. DataMa Pivot might help you in understanding this better.

 

  • Display options (for Shiny app)
    • Contextual help: Display executive summary and contextual worded sentences in help tooltips of charts. Swith off to avoid contextual help in the interface
    • Contextual help precision: Out of 10, this ladder allows to adapt the level of detail available in the contextual help (executive summary, info bubble,…). By default the contextual help is at 5.
    • Print version: allow to read Shiny chart of segment performance without hovering on it
    • Output unit (output_unit): the unit of the overall KPI you’re following. By default it is set to « € » for « Revenue »
    • Dimension excluded from Summary View (Dimension_filter_SimplifiedView) : A dimension you don’t want to consider within summary view, for some reason. The dimension is kept in the calculation but just don’t appear in graphs
    • Label Bridges: Choose whether you want to display numbers on watefall or not

Articles