Configuring Data Intervals

This article shows you how to configure data intervals for your tracked timing and value data in order to have a valuable data visualization.

The Feature Timing and Feature Value data that you track are all individual, discrete values. Each individual values has little value but when you add the appropriate semantics on top of the value, then a valuable visualization of the incoming data is possible.

When you go to configure intervals you will be looking to set up a number of intervals where each tracked value will fit into one of these intervals. By have each tracked value fit into a single interval, it becomes possible to visualize the distribution of values between the intervals. So the goal when configuring intervals is to create intervals that make up semantic groupings of your values between which it would be valuable to see the distribution over time.

Configuring the Intervals

When you have tracked your first timing or value data and try to access it in the analytics client, you'll likely encounter the following:

AnalyticsClientDataConfigureIntervalsNoData

This means that you have yet to define some semantic data intervals and the data cannot be visualized without such intervals configured. You can click the Configure Intervals button (which is also always available in the Menu as well) to begin setting up these intervals.

AnalyticsClientDataConfigureIntervalEmpty

The Configure Intervals dialog consists of a lefthand side where the data intervals are entered as the dividing data values and a right hand side that shows chart of the distribution of the values (based on a set of sample values). The right hand side is generally intended to be an assistance in defining the data intervals by showing the overall distribution and suggesting a few quick-selection options that adds data intervals based on different percentiles in the sample set.

Note that the right hand side of the dialog may be empty if insufficient values are available to generate the sample set for the given feature.

An example

Let's show the configuration process through an example; Let us image tracking the time it took for your application to go from being started to actually showing a workable user interface to the user. This value could be used as an indicator for the first user experience and you might even have your own internal goals as to how fast the UI should show up. Note that the values tracked from you application (through the integrated analytics monitor) will report time in milliseconds but you can change the scale and unit through the Settings dialog.

Based on your internal goals and your overall expectations, you can define a number of intervals that divide the tracked startup time into a number of reasonable intervals that makes sense to you and your organization when reasoning about startup time. For this specific application we determine that we want to divide the values into intervals of: less than a second, between 1 and 3 seconds, between 3 and 10 seconds, between 10 and 25 seconds and above 25 seconds. These intervals can be seen defined in the screenshot below:

AnalyticsClientDataConfigureIntervalExample

Notice the dividing values that are defined on the left hand side and also note how these dividing values are indicated in the sample distribution chart on the right hand side. In this example the values were entered manually using the Add button at the top, but we could also have clicked on any of the quick selection percentile options to generate some dividing values.

You accept the data intervals by clicking the Set Intervals button at the bottom. Changing the intervals will cause the analytics service to re-distribute all the tracked values across the new intervals. This may take some time so you will likely see the following in place of your data:

AnalyticsClientDataConfigureIntervalUpdating

Once the new intervals have been applied the distribution of values across these intervals is available for interaction, including applying filters. An example of such values can be seen in the screenshot below:

AnalyticsClientDataConfigureIntervalNew

You can both consider the overall distribution of the values in the pie chart to the right or you can focus on the distribution over time by looking at the stacked chart to the left.

Tips when Defining Intervals

There are a few things that could be considered when defining your intervals:

  • We recommend that you alter your scale and unit before starting to configure intervals, as you may enter up with slightly odd interval definitions if you scale after you configure the intervals
  • All values must be included in one single data interval. If you have some tracked values that fall outside of your normal range of values (due to e.g. measurement faults, human typing errors, timing issues due to hibernation of a computer) we recommend that you specifically define intervals that includes these values which can then be unselected in the legend to provide a more sane look at the data.
Start a free trial Request a demo
Contact us: +1-888-365-2779
sales@telerik.com
Copyright © 2016-2017, Progress Software Corporation and/or its subsidiaries or affiliates. All Rights Reserved.