Documentation of Datama Studio.Using an earlier version? Previous documentation
Charts
The Tree module of Pivot shows your KPI as a tree, from the most interesting step to the segments that explain it.
The tree
Section titled “The tree”| Level | Shows | Gauge |
|---|---|---|
| KPI | The total of your market equation. | |
| Step | One step of the market equation, with its value. | Step interest: how dispersed the step is across its most interesting dimension. |
| Dimension | A dimension that splits the step, most interesting first. A 🎯 marks the focus dimension of the step. | Coefficient of variation, with a High, Medium or Low interest pill. |
| Segment | One segment of the dimension, with its value on the step. | Value vs. Mean for a ratio, Value vs. Total for a volume. A pill gives the gap to the average, for example vs avg: +12%. |
Hover a segment to see the numerator and the denominator of the step for that segment.
Explore the tree
Section titled “Explore the tree”- Click a step, a dimension or a segment to open or close it. When the equation has a single step, that step is the root and opens on its own.
- Open a segment to split it again by the dimensions not used yet on its path, for example the devices within one country.
- Right-click a segment, then Filter on this, to filter the story on this segment and its path. See Filters.
- Create a What-if scenario from the shortcuts of the chart.
How Pivot scores
Section titled “How Pivot scores”For each step and each dimension, Pivot computes the value of the step for every segment, then the coefficient of variation of the segments: their standard deviation, weighted by their volume, divided by the average. The higher it is, the more the segments differ.
- A dimension with a coefficient of variation of 15% or more is of High interest: its segments really behave differently. At 0.5% or less, it is of Low interest.
- The interest of a step is the highest coefficient of variation among its dimensions. Steps are sorted by interest.
- Small segments are grouped into Other first, following the clustering settings.
The title of the chart says what stands out: a dimension with high discrepancies within a step, no striking dimension, or no real difference between segments.