how to handle scaling of results

[tomjemmett]

in the national notebook we sample data, but we do not record the sample rate used, or scale the results back up. we should do one of these, otherwise we have got to guess what the sample rate used was

[DCEW]

So, just to clarify, the results of the National model run are 1/100th the size 'expected' size? I presume users would expect to see numbers that would refer to the whole National data?
Would it simply be a case of multiplying all aggregate results @tomjemmett ?

[francisbarton]

@tomjemmett @StatsRhian can I get a view on whether it's valid (or "valid enough") for users to simply take values in national results files and multiply them by 100 (as we currently base results on a 1% sample of national activity data)?

[tomjemmett]

I need to re-familiarise myself, but I think the rows are sampled/scaled down to a rate (by default 1%), and the results are not scaled back up after.

so, currently, yes, multiplying back would be fine. but I don't think we record the sample rate, so it would have to be assumed. @yiwen-h, feel free to correct?

[yiwen-h]

The sample rate is specified inside the notebook itself (see cell "Set up the data") but it is not recorded in the metadata, which is Not Great