Core Data Assets Changelog¶
Geographies¶
cbsa
(no changes)
geography_registry
v1 > v2¶
- The column geometry_projection_code has been renamed geometry_projection.
v2 > v3¶
In addition to the changes from v1 to v2, we have:
- Geographies from the whole world instead of only the US.
- Different naming conventions in the partition key geography_type_code: block_group → admin4.
- New types of geographies not related to the usual census division for the US: CSA, CBSA, DMA, and timezone.
- New format to represent the admin4 from
admin1.admin2.admin3.fips_code
toadmin1.admin2.admin3.admin4
(e.g.: US.CA.037.060374082122 → US.CA.037.408212.2). - Extended coastlines: part of the sea has been assigned to the neighboring geographies so as to both simplify them and to be able to assign points in the sea (at least those close to the coast) to a certain country.
- Addition of columns calculated based on GHS data, i.e. the geography population centroid coordinates (centroid_lat, centroid_lng).
- Removal of the following columns: geometry_projection, census_year, geometry_geojson, geometry_wkb.
Please refer to the tutorial on geographies to see practical examples of what has just been mentioned.
POIs¶
brand
public_poi
public_poi_history
custom_poi
custom_poi_history
v1 > v2¶
Significative improvements have been introduced from v1 to v2. Please refer to the tutorial notebook dedicated to the POIs to deep-dive into details.
Here are listed the main points of attention:
- A new structure of
brand
table and enrichment with the sic_code and sic_name columns. - The obsolete
poi
table has been dropped. - The 2 public tables have been renamed into
poi
andpoi_history
tables. - Addition of validity columns in the 2
poi_history
tables (valid_from_timestamp, valid_to_timestamp) made to join with the new field place_version in thevisit
table. - Addition of columns: opening_date, closing_date, distributor_flag (the last just for Spectus POIs).
v2 > v3¶
- No changes.
Device Metrics¶
monthly_distinct_devices
daily_distinct_devices
device_metrics
device_feature
v1 > v2¶
- No changes.
v2 > v3¶
- All missing tables in the v3 schema. This is a temporary solution: the first 3 tables will collapse into one single master table, namely the
device_metrics
. - The table
device_feature
has been dropped.
Home-work¶
hw_with_tag
v1 > v2¶
- No changes.
v2> v3¶
- The table has been renamed into
device_recurring_area
. - A new snapshot of the
hw_with_tag
table is computed weekly, while indevice_recurring_area
it is daily done. - The column processing_week_code has been therefore replaced by the column snapshot_event_date.
- Addition of the columns os_name and confidence_level.
- New partition column provider_id.
- Improved algorithm: please refer to the dedicated tutorial to deep-dive into the matter.
- The table now refers to the new geography table, hence a new notation for the Census block group has been adopted. See the Geography section of this very migration guide for details on how.
Device Location¶
device_location
device_location_uplevelled
v1 > v2¶
- No changes.
v2> v3¶
- Removal of the column lastseen_unixdatetime.
- Removed the IP column from the
device_location_uplevelled
- Addition of the column os_name.
- New partition column provider_id.
- The value 'PERSONAL_AREA' of the classification_type field has been renamed into 'RECURRING_AREA' in the
device_location_uplevelled
table. - The table now refers to the new geography table, hence a new notation for the Census block group has been adopted. See the Geography section of this very migration guide for details on how.
Stops¶
stop
stop_uplevelled
v1 > v2¶
- Addition of the column block_group_id.
v2 > v3¶
In addition to the changes from v1 to v2, we have:
- Addition of the column os_name.
- New partition column provider_id.
- The value 'PERSONAL_AREA' of the classification_type field has been renamed into 'RECURRING_AREA' in the
stop_uplevelled
table.device_location_uplevelled
table. - The table now refers to the new geography table, hence a new notation for the Census block group has been adopted. See the Geography section of this very migration guide for details on how.
Visits¶
visit
v1 > v2¶
- Improved the visit algorithm being used to populate the table.
- Addition of new columns brand_id and place_version.
- Addition of new columns geohash and zipcode_id.
- The start_timestamp field has been renamed into zoned_datetime.
v2> v3¶
In addition to the changes from v1 to v2, we have:
- Addition of the column os_name.
- New partition column provider_id.
- The table now refers to the new geography table, hence a new notation for the Census block group has been adopted. See the Geography section of this very migration guide for details on how.
Socio-demo¶
device_user_labeling
segment_taxonomy
No changes from v1 to v2/v3.
Trips¶
trip
v1 > v2¶
- No changes.
v2 > v3¶
- The table
trip
has been dropped: you'll be able to fetch the same information from thetrajectory
table that will be released soon.