Joining DataFrames (or Series) and Series.

[jseabold]

pd.version.version
'0.12.0-1149-g141e93a'

Maybe already fixed, but I'm afraid to upgrade in the middle of working on a project given past experiences.

As I recall there were several ways proposed to join a series to a DataFrame with join being the preferred way.

df = pd.DataFrame(np.random.random((30,2)), columns=["A","B"])
series = pd.Series(np.random.random(30), name="C")

I know I can do

df["C"] = series

or

df = df.join(series)

But it'd be nice if the following also worked given that I'd like to write general code.

pd.concat((df, series), axis=1) # ok error message at least
df.merge(series) # just call join under the hood? bad error message

This is "unexpected" that it somehow tries to append the rows. Looks like a bug to me. Should it take an axis keyword?

df.append(series)

[jreback]

this would need to be implemented (not difficult), and I agree should be done...

[jreback]

work around in 0.13 (to_frame is new)

def f(x):
   if isintance(x, Series):
        return x.to_frame()
   return x

pd.concat([ f(x) for x in list_of_stuff ])

[jorisvandenbossche]

related #2385?

[jreback]

@jorisvandenbossche definitely...there are a number of condition when this SHOULD raise, but should work otherwise.....need someone to sort them out...

close this issue and put enhancement on the other one?

[jreback]

dup of #2385