Merge pandas dataframe with unequal length

Question

I have two Pandas dataframes that I would like to merge into one. They have unequal length, but contain some of the same information.

Here is the first dataframe:

BOROUGH  TYPE  TCOUNT
  MAN    SPORT   5
  MAN    CONV    3
  MAN    WAGON   2
  BRO    SPORT   2
  BRO    CONV    3

Where column A specifies a location, B a category and C a count.

And the second:

BOROUGH  CAUSE  CCOUNT
  MAN   ALCOHOL   5
  MAN     SIZE    3
  BRO   ALCOHOL   2

Here A is again the same Location as in the other dataframe. But D is another category, and E is the count for D in that location.

What I want (and haven't been able to do) is to get the following:

BOROUGH   TYPE   TCOUNT  CAUSE  CCOUNT
  MAN    SPORT     5    ALCOHOL    5
  MAN    CONV      3      SIZE     3
  MAN    WAGON     2      NaN     NaN
  BRO    SPORT     2    ALCOHOL    2
  BRO    CONV      3      NaN     NaN

"-" can be anything. Preferably a string saying "Nothing". If they default to NaN values, I guess it's just a matter of replacing those with a string.

EDIT:
Output:

<class 'pandas.core.frame.DataFrame'>
Int64Index: 233 entries, 0 to 232
Data columns (total 3 columns):
BOROUGH                          233 non-null object
CONTRIBUTING FACTOR VEHICLE 1    233 non-null object
RCOUNT                           233 non-null int64
dtypes: int64(1), object(2)
memory usage: 7.3+ KB
None

<class 'pandas.core.frame.DataFrame'>
Int64Index: 83 entries, 0 to 82
Data columns (total 3 columns):
BOROUGH                83 non-null object
VEHICLE TYPE CODE 1    83 non-null object
VCOUNT                 83 non-null int64
dtypes: int64(1), object(2)
memory usage: 2.6+ KB
None

Answer 1

Perform a left type merge on columns 'A','B' for the lhs and 'A','D' for the rhs as these are your key columns

In [16]:
df.merge(df1, left_on=['A','B'], right_on=['A','D'], how='left')
​
Out[16]:
   A  B  C   D   E
0  1  1  3   1   5
1  1  2  2   2   3
2  1  3  1 NaN NaN
3  2  1  1   1   2
4  2  2  4 NaN NaN

EDIT

Your question has changed but essentially here you can use combine_first:

In [26]:
merged = df.combine_first(df1)
merged

Out[26]:
  BOROUGH    CAUSE  CCOUNT  TCOUNT   TYPE
0     MAN  ALCOHOL       5       5  SPORT
1     MAN     SIZE       3       3   CONV
2     MAN  ALCOHOL       2       2  WAGON
3     BRO      NaN     NaN       2  SPORT
4     BRO      NaN     NaN       3   CONV

The NaN you see for 'CAUSE' is the string 'NaN', we can use fillna to replace these values:

In [27]:
merged['CAUSE'] = merged['CAUSE'].fillna('Nothing')
merged['CCOUNT'] = merged['CCOUNT'].fillna(0)
merged

Out[27]:
  BOROUGH    CAUSE  CCOUNT  TCOUNT   TYPE
0     MAN  ALCOHOL       5       5  SPORT
1     MAN     SIZE       3       3   CONV
2     MAN  ALCOHOL       2       2  WAGON
3     BRO  Nothing       0       2  SPORT
4     BRO  Nothing       0       3   CONV