Understanding the Standard Deviation Graph - trillium_discovery - trillium_quality - 17.1

Trillium Control Center

Product type
Software
Portfolio
Verify
Product family
Trillium
Product
Trillium > Trillium Discovery
Trillium > Trillium Quality
Version
17.1
Language
English
Product name
Trillium Quality and Discovery
Title
Trillium Control Center
Topic type
Overview
Administration
Configuration
Installation
Reference
How Do I
First publish date
2008

For all numeric attributes, once you run statistical analysis, and then go to view the statistical summary of the attribute, you will see the standard deviation graph. You will also see the standard deviation graph for entities that contain numeric attributes.

The graph above shows the distribution of data and highlights the spread around the mean. All values that are within 1 standard deviation of the mean are plotted in the green band; values that lie between 1 and 2 standard deviations are plotted in the yellow band; values between 2 and 3 standard deviations are plotted in the pink band.

In the graph, the term m indicates the mean of the distribution. The term s indicates standard deviation, so 1s denotes 1 standard deviation from mean, 2s denotes 2 standard deviation from mean, and so on.

The percentage you see when you hover your mouse over a point is the value used to plot that point, which:

  • for points up to and including the peak of the bell are the cumulative percentage of values for the point (e.g. for the point at -1s the value is the percentage of points that have a value <= the value of mean minus one standard deviation; for the point at the peak of the curve the value is the percentage of points that have a value <= the value of mean);
  • for points to the right of the peak of the bell are the cumulative percentage of values for the point subtracted from 100 (e.g. for the point at +1s the value is 100 minus the percentage of points that have a value <= the value of mean plus one standard deviation).

You can manually calculate what percentage each of the plus and minus points should be from the deviations table (by adding up the percentages in the Distribution column). However, to calculate the percentage at the mean point you have to get the mean value from the Entity Member Attribute table, then drill down to Values, sort by value, and then add up the distribution percentages for all values <= mean.

For the chart shown above, here is a breakdown of how the 78.58% was calculated:

Stats Data5: Value Count: 14

Mean: 22.5

Std Dev: 62.177

Values Deviation

-100 -1

• -39.677 (-1)

4 0

4 0

4 0

4 0

4 0

7 0

15 0

18 0

18 0

22 0

• 22.5 Mean (0)

45 0

60 ` 0

• 84.68 (+1)

• 146.85 (+2)

• 209.03 (+3)

210 +3

Points on Bell Curve:

-1σ 7.14% points <=(mean- std dev) = 1 > 1/14 (value count) = .0714

78.58% points <= mean = 11 > 11/14 = .7857

+1σ 7.14% points <= mean+std dev) > 13/14 = .9286 – 100 = .0714

+2σ 7.14% points <= mean+std dev) > 13/14 = .9286 – 100 = .0714

+3σ 7.14% points <= mean+std dev) > 13/14 = .9286 – 100 = .0714