QA Iteration Process

The challenges of research and development for a “Moving Target”

About the QA Iteration Process

This phase of processing was designed to improve the speed and capacity of the PS File Fix stages that result in final product files. It is the ultimate goal of any researcher working with digital products to create a scenario where steps can be batched, and do not require manual intervention. It is the awareness of any experienced researcher, however, that inevitability there will be a phase that involves checking the results of whatever can be batched, and this is generally referred to as quality assurance (QA).

Due to the large nature of the dataset, there were some processes that required development of temporary protocols that were then enacted to test their viability. In the scenario of the QA RR Iteration Process, the first round could only be described as a failure. At this time, the end product of the map series was unknown and it was the decision of the lead researchers to err on the side of caution with high standards for Quality Assurance. The hope was that employing high QA standards would prevent the need to go back and do re-work. However what was unforeseen was that the capacity to fix files improved as the techniques for quality assurance improved, and thus all the blocks processed during the early stages inevitably had to be run through to make sure they were up to standards that were set once a final product was through RnD and the “Bundles” technique was in place.

QA Iteration Process Part 1 – RRs and Early Stages

NOTE: This Part 1 document is included for archival purposes only. If this process is required for future datasets the “Bundles” technique should be employed.

Initially it was thought that tracking established RRs (replace color range values) would allow for certain patterns to emerge, such that files with given characteristics would respond to certain RR values, and a batch process could be applied. Unfortunately, no such patterns emerged, and the hope of batching this stage soon proved impossible.

This process evolved into what eventually became the official process, and many of the techniques and descriptions were carried forward. The associated progress map eventually became colloquially known as “the clown pants map” as a result of the extreme variability in standards between operators, as opposed to a metric for how many files were problematic in the block as was originally intended. Thus, the difference between the “clown pants” region of this choropleth map and the predominantly green and blue areas is more of a distinction between the two stages of development than it is an indicator of where the more or less challenging blocks are located. There is some validity however that the multicolored areas were also the areas of early scanning, and these areas continue to be more challenging, but the map does not necessarily preclude the difficulty of any given block or represent a reliable pattern.

QA Iteration Process Part 2 – Bundles

This process was used for the majority of file sorting, predominantly for all areas west of the 80th parallel (not including Michgan). This process involved a combination of batching and manual file sorting to create “bundles” or sorted folders within each block that would assist the PS File Fix team with increasing their file times and overall speed.

It is important to note, however, that in this phase of creating bundles, the RR legacy continued as there were commonly used RRs employed through the process.

Powered by WishList Member - Membership Software