POWER BI: Modeling

Statistical Functions


CONFIDENCE functions

Confidence functions in DAX are used to help compute confidence intervals about the average value called the sample mean. The value that gets returned is the margin of error (ME ) defined as:

$ME = {CV \sigma \over \sqrt{n}}.$

CV stands for the critical value which is determined based on the significance level alpha that you specify.

For a 95% confidence interval, alpha is set to 0.05

There are two types of CONFIDENCE functions in DAX. The first, CONFIDENCE.NORM(alpha,standard_dev,size) returns the calculation of the margin of error for the normal distribution. The second, returns the margin of error for the student's T distribution:

                  Margin of error = CONFIDENCE.NORM(0.05,10,30)
                
	            

This returns a value of 3.58

                  Margin of Error = CONFIDENCE.T(0.05,10,30)
                 
	            

This returns a value of 3.73

When used with the sample mean, $\bar{x}$, one can specifiy a range of values the true population mean, $\mu$ falls within.

For example suppose the sample mean, $\bar{x}$ has a value of 20. Using the margin error above our confidence interval becomes:

  • $\bar{x} \pm ME$ (formula for calculating confidence interval)
  • $ => 20 \pm 3.58$
  • $ => (16.42,23.58)$

Using the normal approximation, the true population value is between 16.42 and 23.58 19 times out of 20. (ie. 95% of the time).

A Student's T distribution always returns a slightly higher value then the normal distribution and has slightly "fatter tails" which result in a slight higher probability that extreme values will be observed. When you have a large sample size, the two distributions converge. This means the confidence intervals will be the approximately the same.


Dash-Intel is a Power BI and Tableau resource site for data visualization and building BI dashboards.

Data Analyst & Consultant

Copyright 2015-2023 Dash-Intel.com Terms