Introducing Historical Trend Reporting in Jazz Reporting Service Report Builder 6.0.1

Steven Shaw
Last updated: April 11, 2017
Build basis: Jazz Reporting Service Report Builder 6.0.1

The Report Builder now supports creation of historical trend reports. These reports display trends of certain measures over time, possibly separated by an attribute dimension. Their unique character warrants a different flow than reports on current data. We have introduced a new section where you can choose your report type. You can pick either Current Data (same as existing flow) or the new Historical Trends report type, which has, instead of the traceability section, a section where you set the time range for the trend you are measuring. Historical trend reporting is currently available only for the Rational Data Warehouse data source.

Note: For a continued discussion of trend and aging reports in version 6.0.3, see Trend and Aging Reports in Jazz Reporting Service Report Builder 6.0.3.

Single trend with dimensions

Let’s consider an example where you want to display the total work items separated by a value dimension of their status (resolved versus unresolved). First, you pick the Historical Trends report type:

Next, you decide what trend and artifact you are interested in measuring. In the Focus on artifact trends section you select from a list of artifacts and associated trends. Since you want to measure work item totals, you pick the basic Totals trend under Work Item. Alternatively, if you wanted to find out about work item creation trends instead, you’d pick Creation trend. For information about specific trends refer to the appendix section of this document.

The next step is to select a time range to capture the trend within. In the Set a time range section the default time range is set to the last 3 months, and you can change it to any range you like. For this example, let’s set it to the last 2 months.

It’s also possible to set specific dates using the Calendar drop-down or set an End date to a relative date in the past (i.e. 14 days ago). You’ll notice under Conditions that you now see new conditions for Start date and End date after you set the time range, and you can change them here, too.

Continuing on to Format results, you’ll notice the default is a graph presentation. Historical trend reports are typically shown as a line graph, but you can also use a bar chart (stacked or grouped).

In the graph section you need to select the unit to measure: the total count of the related artifact (work items) or some other criteria that was captured in the metrics artifact. You can also select a dimension to separate the value being measured by all the values of the dimension, such as Status, Filed Against and so on. Before you can preview a graph, you must select a value for the unit of measure, and optionally, the dimension.

After you select Total Work Items for the unit of measure and Status Group Category for the dimension, the graph should look something like this:

Then you can modify the line colours, change the date scale to weeks or months or alternatively, view the data in a bar chart. Now the report is done, and you can save and publish it to a dashboard like any other report created by the Report Builder.

Multiple historical trends 

The ability to report on a historical trend capability is based on selecting a specific trend, such as creation or totals, and seeing its change over time. The artifact type and trends are organized hierarchically, you can look for the artifacts you want to report on and expand to see what trends are available.

Many different trends are stored in the data warehouse as part of the data mart / fact collection process in the Data Collection Component (DCC). For instance, work items have trends for discovering creation rates, general totals, and resolution rates. Now you can visualize more then one trend on a graph so you can compare them. You can select more than one trend in the Focus on artifact trends section:

Setting the time range is the same as is the single trend scenario. The time range will be effective for all selected trends.

Setting Conditions

Setting conditions that restrict or constrain the data is identical to how you would do this for a current data report (i.e. on a Work Item). Though the set of attributes available is limited to a specific set that were captured in the trend table. One limitation is that custom attributes are not available on the trend artifacts as well as some other standard attributes (Tags).

To add conditions, launch the conditions dialog and select the trend you want to add a condition for, followed by choosing an attribute and which values to constrain it to. For instance if we want to constrain our report to a particular release and for defects only, then we need to add Planned For and Type conditions for each trend we’ve selected (i.e. Creation, Resolution and Totals). After doing this, our Set conditions pane should look something like this:

Then we need to consider if there are any special considerations we need to make on each trend (see appendix for details). As an example, when considering total work items, we are probably only interested in currently open ones. Especially considering we’re looking at the resolution trends at the same time. We can add this by selecting the Totals trend for Work Item in the Attributes of drop down in the condition dialog and selecting the Status Group Category attribute and constraining it to unresolved.

The Resolution trend captures metrics around total number of resolved and closed Work Items on a given day. Consequently if we’re only interested in straight resolutions and not closures then we need to add an additional condition there on the State attribute to check only for resolved. Creation trend captures all Work Items created on a certain day so no special considerations need to done there.

Limiting or excluding trend dates

For trend reporting, you can select a date range to restrict the results. However you might want to restrict the result further by only allowing work days (i.e. Monday to Friday) or specific months or quarters of the year. You can do this when adding a condition by selecting the Date artifact type.

This ability could be most useful for improving report performance by reducing the results to a specific day of the week. If you’re examining a long date range, you can pick a specific day of the week to draw the trend from (such as Wednesday), significantly reducing the number of results the query and chart engine have to process while providing the same trending information.

Adding a condition on the Date artifact affects all the trends though. This might not be desirable if you need to accumulate trend values over the course of a period of time though. Especially if the trend captures some sort of daily value that may vary from day to day. In our example, we have one trend that captures a consistent snapshot – Totals. The total from one day to the next is stand-alone and doesn’t depend on the previous day. It can be compared to a total from the previous week if desired. So adding a Day of Week condition makes sense. For other trends such as creation or resolution, the metrics in these trends are a daily total – i.e. How many work items were created or resolved on a specific day. If you want to see a weekly trend total for these you can’t exclude any particular days from the result. So to add a Day of Week date condition on a particular trend, we need to select the trend in question and look for the Day of Week attribute.

Now we can constrain only that particular trend to one or more days of the week without affecting the other trends data.

Formatting the trend data

Then, in the Format Results section, since we want to visualize multiple measures, you select the radio button to add lines or bar segments based on numeric measures. This allows selection of more then one measure from one or more trends. Each of the selected measures can be displayed as a line or a bar segment, depending on which graph type you are using.

After you click Add measures you can choose an available trend and pick a measure to visualize in the graph. In this case you are interested in the Total Work Items measure that provides a count of all work items filtered by conditions you’ve added previously on the Choose Data tab. You can add multiple measures at one time by flipping between the different trends in the Attributes of drop down. For this example flip to the resolution and creation trends and select the Total measure for each of them.

After you choose the measures, the graph preview is generated with a line representing each measure. You can also choose the type of graph and the data scale (such as Weeks or Months) for the X axis. In this example we will change the date scale to be Weeks and modify the graph type to use Grouped Bar so we can easily differentiate and compare the measures on a week to week basis.

You might also want to decide how the data is gathered within the date scale period. the default is to average all of the data points discovered within the date scale range. You can also sum all the date points or consider only the first or last data points. In this example you want to sum all the created and resolved work items totals for the week since we want a total for them by week. For the Total Work Items measure we can leave it as is since we’ve already filtered it in the Conditions pane to a single day of the week. As such, the calculation is irrelevant for it since there will only ever be one value in each week interval to consider.

To polish the results, you may want to modify the default colors to something specific or make certain lines stand out versus the others. A drop-down control in a table underneath the Add measures button allows the color of an individual line or bar to be changed to a pre-defined color or you can even pick a particular RGB value. It is also possible to rename the legend entries to be more concise or consistent with the other entires. In the example we’ve changed the colors to red, green and blue and modified the labels to be more readable.

Set a Goal line

If you want to track a trend to a particular goal, click Add goal line. A dialog box opens so you can select a particular value.

Your value will display as a horizontal dotted line in your graph. You can change its color and add additional goal lines by repeating the process. The legend is sorted alphabetically right now so if you want to adjust the order of an entry you’ll need to change it’s label accordingly. In the example, we’ve modified the label of the goal line to be “A Goal (27)” so that it appears first in the legend.

Validation for historical trend reports

Historical trend reports query against a vast set of data and potentially across multiple project areas.  Some reports can take a long time to run if they are not filtered to a smaller subset of data. To make this easier, we have added validation to ensure that some constraints are in place to increase performance.

1. A time range is mandatory when creating trend reports. Although an infinite range is still possible, it is better to set a range within a specific time frame. You cannot format the results until you enter a time range.

2. The graph must include a value for the unit of measure. Again, this helps the report generation optimize the set of tables to query against.

3. You get a warning when you save if no conditions and project scoping are set. It is advisable to set some scoping to narrow down the results, such as work items with conditions on the Release (Planned for) and / or category (Filed Against) fields.

Appendix

Change and Configuration Trend Details

Quality Management Trend Details

Requirements Management Trend Details

Copyright © 2016 – IBM Corporation