search

brittanyblouin / ANCRTAdjust

An R package to adjust routine HIV testing data from antenatal care to reduce bias in estimating HIV prevalence trends
MIT License
2 stars 3 forks source link

Quality indicators by time #9

Open ellessenne opened 4 years ago

ellessenne commented 4 years ago

Reporting quality indicators by time would be more effective by plotting rather than printing the results to the R console - in the example it is a very tall dataset! I would suggest adding a function to do so, maybe it could accept a "quality_indicators" object as input and could be used for plotting quality indicators by region as well. Does it make any sense?

seabbs commented 4 years ago

Second this. Could use S3 and give it a plot method.

For this #openjournals/joss-reviews/issues/1740 review

m-maheu-giroux commented 4 years ago

We already have summary of data indicators that are plotted by time in the descriptive_plot() function. I feel that we would be duplicating efforts if we were to plot this twice. The objective of this function was to provide a tabular summary of the quality indicators. In most instances, users will want to have this summary for the whole ANC-RT dataset.

I agree that if the users stratify by subnational unit and by time, the output database becomes very lengthy... but in almost all instances, I expect that users will run this function on the whole dataset.

ellessenne commented 4 years ago

I think this could be ok, but the output of quality_indicators is a data set with string columns - this makes impossible for users to create the above-mentioned plot themselves.

Using the example dataset (ancrt), I get a dataset with 976 rows for different combinations of region/time - this feels quite cumbersome for the user to go through. I understand that you might expect users to compute indicators for the overall dataset (with no stratification), but wouldn't this be dangerous? I mean, shouldn't quality be assessed in great detail?

What do you think @seabbs?