Who is this article for?

Users and Administrators who want to understand trending

A Trend Analysis license allows use of the trend output to analyze trends over time. Depending on your organization's contract, Trend Analysis may not be enabled in your subscriber area. Contact your Technical Account Manager for additional information.

Trending is one of the many available BI visualization options. Create a Trend Analysis report by searching for data and selecting Trend Analysis from the Custom Visualization options.

Mean and Standard Deviation Calculations

The Mean and Standard Deviations used in the scoring profiles (below) are derived from the data in the Baseline Period. The Mean is the sum of all the baseline values divided by the number of baseline values. The Standard Deviation is the square root of the Variance, which is defined as the sum of the squares of each values distance from the mean, divided by the number of data points minus 1.

So given 12 values in a baseline period of 5, 11, 12, 3, 0, 1, 15, 10, 9, 5, 6 and 8, we would calculate:

Mean

Variance

Standard Deviation

References to Standard Deviation in Trend Profiles will typically refer to the number of Standard Deviations a value is from the mean. So "more than two standard deviations below the mean" would mean that a value was more than two times the standard deviation amount below the mean value. In the example above, that would mean anything below -2.04 (4.56 x 2 = 9.12; 7.08 - 9.12 = -2.04) and none of the data points would qualify.

Control Limit Calculation

Control limits are automatically calculated and applied to any relevant rules associated with the current trend profile. The basis for the control limit calculations is the Shewhart individuals control chart. The system performs the following calculations:

• The average (x̄) of the occurrences for each period within the baseline range.

• The average Moving Range (MR) over the baseline range where the moving range is defined as the absolute value of the difference between the number of occurrences in period(n) and period(n-1).

• The Upper and Lower Control Limits (UCL & LCL) are a constant value derived from the x̄ and the Average Moving Range from step 1 and 2. These values are calculated on the Baseline Period data set and applied to the Trending Period. Assuming the Trending Period and Baseline Period do not overlap, no data within the Trending Period influences the UCL and LCL.

Scoring Criteria

Each of the following scoring criteria, based on Nelson Rules, may be applied to a trend profile. There are two versions of each rule, one that considers data offset from the most recent data point and another that considers all the data within the Trending Period range. For example, the rule "The most recent X points are alternating in direction" considers the most recent X points starting with the current period. The corresponding floating rule, "X points in a row are alternating in direction", considers whether any X points in a row within the Trending Period range are alternating. In the examples below, red data points indicate points that violate the rule.

Ensure the Trending Period includes the needed periods
If the selected trending profile contains scoring criteria that define several periods, such as "The most recent X out of Y points..." the Trending Period must be large enough to include these periods. If the Trending Period is restricted to a duration that doesn't include all needed periods, the scoring criteria are ignored.

Rules and Examples The most recent X out of Y points are above the Upper Control Limit 1 Point is above the UCL

The most recent X out of Y points are below the Lower Control Limit 1 Point is below the LCL

The most recent X out of Y points are above the mean 4 out of the last 5 points are above the mean

The most recent X out of Y points are below the mean 2 out of the last 4 points are below the mean

The most recent X points are increasing The 4 most recent points are increasing

The most recent X points are decreasing The 4 most recent points are decreasing

The most recent X points are alternating in direction The three most recent points are alternating

The most recent X out of Y points are more than one standard deviation above the mean 3 out of the last 5 points are more than one standard deviation above the mean

The most recent X out of Y points are more than two standard deviations above the mean 2 out of the last 4 points are more than two standard deviations above the mean

The most recent X out of Y points are more than three standard deviations above the mean 3 out of the last 5 points are more than three standard deviations above the mean

The most recent X out of Y points are more than one standard deviation below the mean 2 out of the last 2 points are more than one standard deviation below the mean

The most recent X out of Y points are more than two standard deviations below the mean 2 out of the last 2 points are more than two standard deviations below the mean

The most recent X out of Y points are more than three standard deviations below the mean 2 out of the last 2 points are more than three standard deviations below the mean

The most recent X out of Y points are within one standard deviation of the mean (on both sides) 7 out of the last 8 points are within one standard deviation around the mean

The most recent X out of Y points are beyond one standard deviation from the mean (on both sides) 7 out of the last 8 points are beyond one standard deviation around the mean

The most recent X out of Y points are outliers Two points are within the inner fences of the box plot

The most recent X out of Y points are extreme outliers One point is outside of the outer fences of the box plot

The number of occurrences is greater than or equal to X in a set of the most recent Y periods The number of occurrences in the last two periods (Y) is greater than 30 (X)