box plot chart outliers

[Former Member]

In Webi one can create a Box Plot Chart.

A Box Plot Chart (also called a stock chart) displays a five-number summary based on the distribution of a dataset: the maximum, the minimum, the first quartile, the third quartile, and the median.

A Box Plot Chart can also show abnormal values called outliers.

Question: How does Webi decide what values are outliers?

In "Format Chart" option in Webi on a Box Plot, one can tick/untick the option "hide outliers"? What algoritm is behind it?

How does Webi decide what values (outliers) are to be hidden?

[pascal_gaulin · Accepted solution]

Hello Jan,

Outliers are values which are very far from the other box plot values. In order to determine which values are "very far", WebI calculates the inter-quartile range (IRQ): 3rd quartile - 1st quartile.

If a data value is beyond 1.5 * IRQ from Q3, then it is considered as an outlier.

Example:

1st quartile = 27.5

3rd quartile = 47.5

The IRQ is therefore 47.5 - 27.5 = 20

If a data value is bigger than 47.5 + 1.5 * 20 = 77.5, then it is considered as an outlier.

Regards,

Pascal.

[Former Member · Accepted solution]

Thanks a lot Pascal for your useful information.

Is there any way a Webi report developer can modify this "very far" calculation?

[pascal_gaulin · Accepted solution]

Hello Jan,

Nothing I am aware off. However, this parameter could be exposed I suppose, but only if we have enough customers interested of course. If you feel this feature is important for your business, please suggest it on SAP Idea Place: SAP BusinessObjects Web Intelligence: Home

Thanks,

    Pascal.