empty = pd.DataFrame(columns=["x"])
df1 = pd.DataFrame({"x":[1],"a":[3]})
df2 = pd.DataFrame({"x":[1],"b":[6]})
I'd like to merge the df1 with empty.
x a
0 1 3
I tried
empty.merge(df1, on=["x"])
but this returns an empty data frame. After successfully merging with df1, I'd like to be able to merge empty again with df2 so that it results in
x a b
0 1 3 6
If you don't specify 'how' in a merge function it defaults to an 'inner' join which HAS to match keys on both sides of the merge. If you change it to a 'left' or 'right' join when you merge it will preserve the data:
empty = pd.DataFrame(columns=["x"])
df1 = pd.DataFrame({"x":[1],"a":[3]})
df2 = pd.DataFrame({"x":[1],"b":[6]})
df3 = empty.merge(df1, on=["x"],how='right').merge(df2,how='right')
This is clear in the documentation: merge defaults to an inner join. Since there are no keys in your left table, you get no keys in the result. Fix this simply by specifying an outer join:
empty.merge(df1, how="inner", on=["x"])
Full demonstration:
>>> accum = empty.merge(df1, "outer")
>>> accum
x a
0 1 3
>>> accum = accum.merge(df2, "outer")
>>> accum
x a b
0 1 3 6
>>>
