Overview: content on overview page

[pindec]

From issue #17, the "contracts values" chart (of contract values split by value range in USD, e.g n values 0-$10,000) on the overview page is a placeholder.

This is an issue to discuss what content is helpful on the overview page, i.e. candidates to replace that chart and other possible additions.

A few suggestions:

• a chart of objects per stage over time (by year) - as an analyst, I want to see objects published over time so that I understand how a publisher's content has evolved over time with respect to publishing across all stages of procurement.
• a clickable list of failing checks - as an analyst, I want to see an overview of how many checks were failed, and the ability to quickly jump to the relevant check information, so that I can get a quick impression of potential problem areas in the data and the ability to delve deeper.

[jpmckinney]

a clickable list of failing checks - as an analyst, I want to see an overview of how many checks were failed, and the ability to quickly jump to the relevant check information, so that I can get a quick impression of potential problem areas in the data and the ability to delve deeper.

Good idea!

a chart of objects per stage over time (by year) - as an analyst, I want to see objects published over time so that I understand how a publisher's content has evolved over time with respect to publishing across all stages of procurement.

If we want to compare two collections of one source at different times, to see the evolution of the publication, that can be explored in issue #11.

I suppose we can do the same within one collection (as suggested), by comparing compiled releases with dates in different years, but this assumes a publisher never backfills earlier contracting processes (e.g. by connecting a new backend system that has additional historical data). I think the methodology quickly gets messy and unpredictable. For example, a multi-year framework agreement might start in 2015 and end in 2019; its compiled release will be dated 2019, but its quality might be characteristic of 2015 (or a mix).

The only way to draw a clean line between data from 2015 and data from 2019 is to actually collect data from each year – rather than to use the compiled release's date as a proxy.

[jpmckinney]

To scope this issue a bit, and to avoid disappointing any new suggestions along the lines below (copying and expanding on a comment from #17):

The DQT is focused on data's intrinsic quality. As such, the content on the overview page should support an interrogation of the dataset's quality, rather than an exploration of the data it contains. A separate data exploration or business intelligence tool would allow for flexibly slicing data by year or by month, aggregating or disaggregating values based on currencies, changing chart types, etc.

This focus is admittedly blurred by two of the visualizations currently on the overview page.

I also mention intrinsic quality, because there are many extrinsic qualities (like the proportion of government procurement covered by the data) that the tool can't calculate. That said, if there is something that is straight-forward for the DQT to report to support an extrinsic quality metric (e.g. reporting the number of contracting processes), then it can be included.

[jpmckinney]

Added proposals to #50 and #11. Nothing remains, so closing.