使用组合键合并 pandas 中的数据框

Question

I have two dataframes that I need to combine together based on a key (an 'incident number'). The key, however, is repeated, as the database they will be ingested by requires a particular format for coordinates. How can join the necessary columns based on a combination of keys?

For example, the two tables look like:

Incident_Number	Lat/Long	GPSCoordinates
AB123	Lat	32.123
AB123	Long	120.123
CD321	Lat	31.321
CD321	Long	121.321

and...

Incident_Number	Lat/Long	GeoCodeCoordinates
AB123	Lat	35.123
AB123	Long	125.123
CD321	Lat	36.321
CD321	Long	126.321

And I need to get to...

IncidentNumber	Lat/Long	GPSCoordinates	GeoCodeCoordinates
AB123	Lat	32.123	35.123
AB123	Long	120.123	125.123
CD321	Lat	31.321	36.321
CD321	Long	121.321	126.321

The number of records are not 100% equal in each table so it needs to allow for NaNs. I am essentially trying to add the column 'GeoCodeCoordinates' to the other dataframe on a combination of 'Incident Number' and 'Lat/Long', so it will treat the value 'AB123 + Lat' and 'AB123 + Long' as a single key. Can this be specified within code, or does a new column and a calculation to create that value as a key need to be created?

I imagine I went about this in a bit of a goofy way. The Lat and Long were originally stored in separate fields and I used.melt() to make the data longer. The database that will ultimately take this in requires the longer format for the Lat/Long field.

GPSColList = list(GPSRecords.columns)

GPSColList.remove('Latitude')

GPSList.remove('Longitude')

GPSMelt = GPSRecords.melt(id_vars=GPSColList, value_vars=['Latitude', 'Longitude'], var_name='Lat/Long', value_name="GPSCoordinates")

As the two sets of coordinates were in separate fields I created two dataframes with each set of coordinates and melted them separately. My attempt to merge them looks like:

mergeMelt = pd.merge(GPSMelt, GeoCodeMelt[["GeoCodeCoordinates"]], on=['Incident_Number', 'Lat/Long'])

Result is KeyError: 'Incident_Number'

Answer 1

Try:

cols = ['Incident_Number', 'Lat/Long', 'GeoCodeCoordinates']
mergeMelt = GPSMelt.merge(GeoCodeMelt[cols], on=cols[:-1])

The KeyError: 'Incident_Number' is raised because you use GeoCodeMelt[['GeoCodeCoordinates']] so your columns Incident_Number and Lat/Long don't exist when you merge.