Unknown timezones in latest optd_por_public about opentraveldata

[da115115]

Added a file, contributed by @aseredyn:
optd_city_timezone-2016-05-02-AS.xlsx

Some context:

• There are cases of different time zone references for the same city. For instance, Odessa airport (ODS, “A”) is in a different time-zone than Odessa city (ODS, “C”).
• There are more cases, referenced in the above Excel sheet (there are only cases with env_id=="" && !is.na(page_rank) && !is.na(gmt_offset)).
• There are plenty other cases, where some bus station or outdated airport in a given city has some other (or missing) time-zone info than the main one in the city. Those are ignored in this quick check.
• In some cases, the time-zone label is different (e.g., "A, VSA, America/Merida" vs. "C, VSA, America/Mexico_City"). But, in some other cases, the time-zone (and hence GMT offset) is different (like in the case of ODS). In the Excel, those are marked in the “diff_offset” column.
• The numbers of scheduled flights of full 2015 are also provided, so as to sort by the most relevant ones.

[da115115]

Thanks!

Could you please provide a pointer to your script?

[alexprengere]

For more context see opentraveldata/optd#31, the actual script is here.

[tadhgpearson]

+1 for this, it's breaking our timezone import.
I'd encourage you to start by removing the clearly fake places, such as HDQ, VVV, NNN and IAT with no timezone, let me know if there's an obvious way I can help you with that.

[da115115]

There are still a lot of POR with time-zone issues. Almost all of them are now reported in a new CSV file: opentraveldata/optd_por_tz_wrong.csv. However, none of them appear in flight schedules (their PageRank values are null); so, if you filter the null-PageRank POR, you shouldn't have any issue any more.

Those POR correspond to entries appearing in reference data, but not in OpenTravelData.

1. Some of the POR are missing, and should be added to OpenTravelData. For that, add the corresponding entries in opentraveldata/optd_por_best_known_so_far.csv.
2. Some others are invalid, and must be filtered out. For that, just add entries in opentraveldata/optd_por_ref_exceptions.csv.

[tadhgpearson]

Thanks for the clarification Denis. However, it's still not clear to me - do all cities / airports in the optd_por_public file require a timezone? Should we be adding those that don't have a timezone to the exceptions list so that they are not included in the output?

[da115115]

There is no such thing as a file "requiring" a valid time-zone; whereas you, as a user, may require a valid time-zone. If that is the case (and I fully understand that it would be the case!), then a quick work around is to filter out the records having no PageRank value.
On the longer term, we should add to Geonames the missing POR. Then, once the POR are in Geonames, the time-zone will automatically be valid.

[da115115]

All the time-zone errors have been fixed.
See also opentraveldata/optd_por_tz_wrong.csv.

[alexprengere]

I confirm that all timezones are now valid! Thanks Denis!