Layer comparison

The layer comparison functionality allows you to quickly compare tags of interest for different periods of concern, in a numerical and visual way. You will need to create and overlay different layers or periods in order to perform a numerical layer comparison.

In practice the layer comparison functionality can be used to:

  1. Determine root causes by overlaying good and bad periods, numerically compare layers and browse for differences.
  2. Gain process insights by comparing different regimes of operation.
  3. Quickly identify periods with the highest throughput, efficiency, ...

This article describes how to use the layer comparison table and the different statistical parameters available for layer comparison.

How to use the layer comparison table

  • Load in all relevant tags you want to analyze.
  • Navigate to the time period you want to analyze.
  • Overlay other time periods you want to compare.
  • Open the tags menu .
  • Click the arrow icon at the top right of the tags menu to expand the layer comparison table.
  • Click the hamburger icon at the top right of the layer comparison table to select the statistical parameter of interest.

Screen_Shot_2018-07-31_at_10.19.22.png

 

Statistical parameters

Different measurements for performing a layer comparison are available. The 'Average values' option is selected by default. The different measurements can be divided between key statistics and similarity measurements.

An overview of all key statistics available are listed below:

  • 'Average values': each table cell displays the average value of the corresponding tag (row) within the corresponding layer (column). Additionally, the difference with respect to the reference layer (light grey column) is displayed.
  • 'Minimum values': each table cell displays the minimum value of the corresponding tag (row) within the corresponding layer (column). Additionally, the difference with respect to the reference layer (light grey column) is displayed.
  • 'Maximum values': each table cell displays the maximum value of the corresponding tag (row) within the corresponding layer (column). Additionally, the difference with respect to the reference layer (light grey column) is displayed.
  • 'Range values': each table cell displays the difference of the maximum and minimum value of the corresponding tag (row) within the corresponding layer (column). Additionally, the difference with respect to the reference layer (light grey column) is displayed.
  • 'Start values': each table cell displays the start value of the corresponding tag (row) within the corresponding layer (column). Additionally, the difference with respect to the reference layer (light grey column) is displayed.
  • 'Stop values': each table cell displays the stop value of the corresponding tag (row) within the corresponding layer (column). Additionally, the difference with respect to the reference layer (light grey column) is displayed.
  • 'Delta values': each table cell displays the difference between the stop and start value of the corresponding tag (row) within the corresponding layer (column). Additionally, the difference with respect to the reference layer (light grey column) is displayed.
  • 'Integral values': each table cell displays the integral value of the corresponding tag (row) within the corresponding layer (column), supposing that the unit of the tag is units/day. Additionally, the difference with respect to the reference layer (light grey column) is displayed.
  • 'Standard deviation' calculation of the standard deviation over the entire time frame per tag.

An overview of all similarity measures available are listed below:

  • 'Statistical similarity': each table cell displays a measure of similarity between; A) the distribution of values of the corresponding tag (row) within the corresponding layer (column) and B) the distribution of values of the same tag (row) within the reference layer (light grey column). The similarity score is based on an adapted Kolmogorov–Smirnov test statistic.
  • 'Evolution similarity': each table cell displays a measure of similarity between; A) the evolution of the corresponding tag (row) within the corresponding layer (column) and B) the evolution of the same tag (row) within the reference layer (light grey column).
  • 'Correlation': each table cell displays the Pearson correlation coefficient between; A) the evolution of the corresponding tag (row) within the corresponding layer (column) and B) the evolution of the same tag (row) within the reference layer (light grey column).

For key statistics, the difference with respect to the reference layer (light grey column) is displayed in each cell. Positive differences are shown in light orange and negative differences in light blue. The largest positive and negative deviation is underlined for each tag. To change the reference layer, simply click on the header of the corresponding column. To sort based on the values in a given column, click on the arrow up/down icons in the corresponding column header.

Screen_Shot_2018-07-31_at_10.21.50.png

Key statistics are displayed for the reference layer (light grey column), as well as the other layers. In contrast, similarity measures can only be displayed for non-reference layers.

Screen_Shot_2018-07-31_at_10.22.48.png

From version 2017.R2, the Layer Comparison Table also takes into consideration filtered out periods. Whenever a filtered out period is present a yellow label is shown at the top of the table to indicate that these filtered out periods are taken into account.

Screen_Shot_2018-07-31_at_10.24.33.png

Note: Ad hoc filters are considered as one filter. This means that different periods filtered out using the ad hoc filter will be counted as one.

Note: The use of filters are currently not available for similarity and correlation compare options.

Export comparison table

The export functionality of the comparison table can be used to extract the statistical values from TrendMiner to conduct further calculations in tools such as Excel.

Click the export icon   located top right of the layer comparison table to download all statistical parameters of the tags and layers which are visualized in the comparison table.

Layer comparison deep dive video

-- If your TrendMiner screens do not match the screens you see above, this means you probably run an older version of TrendMiner. Contact your administrator for upgrade options.