Business Intelligence: Statistical Methods

Table of Contents 

• Overview
• Percentile
• Percentile Trim
• Trimmed Avg 25:25
• Arithmetic Mean

Overview 

In BT BI, there are now additional statistical means that can be used on BI dashboards and looks. In addition to standard percentile and arithmetic means, percentile trimmed and arithmetic trimmed are now method options. 

Percentile

A percentile represents the value below which a certain percentage of measurements falls. Everything above the configured percentile will be ignored.

When you select Percentile as your Statistical Method, percentile will default 75%, upper and lower filters will say N/A. You cannot use the upper and lower filters with standard percentile. For this functionality, switch to Percentile Trimmed. 

Percentile can configured within the dashboard and can be set 5-99. After selecting a new percentile for the dashboard, hit the update button in the top right to rerun the dashboard. 

Example: 

Percentile Trim

When Percentile Trimmed is selected the default will be 25% upper and 25% lower. This is configured dynamically in the portal. Percentile will show N/A. 

When using this statistical method, the aggregation runs first, then trim occurs in the process. This method is trimmed aggregated data, it is not trimming the raw data. When a trim percentile is selected and the dashboard is updated, the new percentile is calculated using our algorithm. Data accuracy is decreased when using this methodology, but there is increased flexibility. 

Example: 

Trimmed Avg 25:25

When using this statistical method in BI, no values (upper, lower, and percentile) filled by default. This option is a fixed setting of trimming the top and bottom 25%.  This method is recommend when using average to remove outliers.

This method does trim the raw data prior to aggregation.

Example: 

Arithmetic Mean

The arithmetic mean (average) is the sum of all the numbers in the series divided by the count of all numbers in the series.

This option does not trim the data, the entire selected data set will be used to calculate the dashboard. No values (upper, lower, and percentile) filled by default. When using arithmetic mean, data is likely to be skewed by outliers. If this is a concern, please use the Arithmetic Trimmed statistical method. 

Example: