DevonWay Blog

Deriving Insights Across All Your Data With Universal Trending - DevonWay

Written by Wade Watts | July 27, 2020

Trending is a powerful method of data analysis, used to detect anomalies and deviations in large sets of data in an easy-to-consume way. Unfortunately, many organizations still rely on spreadsheets to manage and run trend reports, a process which is tedious, error-prone, and tied to very specific types of data.

DevonWay Universal Trending addresses this market gap by providing an easy-to-use web interface that allows non-technical users to create and run trend reports across any type of structured data: not just trend code libraries, but departments, teams, locations – anything with discrete data values. And since the trends run on top of the DevonWay in-memory NoSQL datastore, results return within seconds, regardless of the number of underlying records.

With DevonWay Universal Trending, users can drill down into a graphical overview that includes a full historical view of all discrete values – or drill down even further into the underlying data to get to the bottom of any deviation.

The tool currently includes 3 standard trend profiles, but with the right permissions, you can create as many as you want. Each profile determines the evaluation criteria that applies coloring to your reports, so you can quickly spot deviations:

1. Heat Map – By Row

The cells of each row in the trend output grid are colored with a gradient between white and red based on their relative size. Administrators can update the coloring scheme for this profile and optionally choose to color the cells based on their relative size amongst all rows.

2. Outliers

Each cell is colored either white, yellow, or red based on whether the value of the cell is either an extreme outlier (red), an outlier (yellow), or normal (white). Outlier detection is based on the simple box-plot approach, where extreme outliers are cells that are beyond one of the outer fences of the baseline data, and outliers are cells between the inner and outer fences. See also, Tukey’s Fences.

3. Statistical Process Control (SPC)

The SPC profile uses the Shewhart individuals control chart approach to calculate whether a cell’s value is in control or not. The SPC profile uses rules to assign violation points based on conditions that may apply to a cell.

Next Platform Release: More Ways to Visualize Data

In the August 2020 platform release, we’ll be expanding on the above profiles to include more customized options:

  • Heat Map – Full Grid: Similar to “Heat Map – By Row,” but the color gradient is calculated using the full grid, meaning every cell from all rows.
  • SPC – Variation Tolerance High: This will allow you to specify a high variation tolerance so that the warning signs are more lenient.
  • SPC – Variation Tolerance Low: Specify a low variation tolerance so that you’ll be warned instantly when a desired trend deviates by a small amount.
  • SPC – Desired Trend Downward: This will allow you to specify that trending downward is favorable. The data in this case will be highlighted if it trends upward so that you can easily spot issues.
  • SPC – Desired Trend Upward: Specify that trending upward is favorable, so your data will be highlighted when it trends downward.

Watch it in Action

To view a trend reporting demo, please contact Us.