PThie commented 2 weeks ago

At grid level data, somehow not everything is anonymized given the 5/50 rule.

Only affects the combined index (I guess).

PThie commented 1 week ago

For CombInd_v13

There is sometimes the value of the price index anonymized but the NOBS is present.
Sometimes the value is not anonymized at all even if it is just 1 NOBS.

The first issue seems to only presented in the change data. The second issue is present in both change and level data (regional effects).

The second issue was intentional as I did not want to anonymize the combined values. However, as the values are not reliable with just one observation, also censor them.

For grid level data (regional effects)

The same applies

PThie commented 1 week ago

This could also be a test which checks if pindex is NA so should be NOBS

PThie commented 1 week ago

Another test:

There should not be a NOBS below the NOBS threshold set (PUF = 50, SUF = 5) If NOBS threshold is undercut, there should be no pindex and no NOBS

PThie / RWI-GEO-REDX

Fix anonymization for combined index #30

For CombInd_v13

For grid level data (regional effects)