pandas conditional logic with mixed dtypes

Question

I have a pandas dataframe as follows:

foo bar
a   b
1   10
2   25
3   9

I want to add a new column as follows:

foo bar baz
a   b   0
1   10  1
2   25  1
3   9   1

Which is: if row['foo'] or row['bar] is numeric, then row['baz'] = 1 else 0

What I have so far is:

def some_function(row):
   if row['foo']>=0 or row['bar']>=0:
      return 1
   return 0

df['baz'] = df.apply(lambda row: some_function(row), axis=1

But this doesn't work because the dtype is not int. I can't drop non-int rows, because I need them in the dataframe.

Any idea how I can solve this?

Answer 1

If want check numeric saved as strings use to_numeric, then compare with ge (>=) and use all for check if all values are True per rows:

df['baz'] = df.apply(pd.to_numeric, errors='coerce').ge(0).all(1).astype(int)
print (df)
  foo bar  baz
0   a   b    0
1   1  10    1
2   2  25    1
3   3   9    1

Or if need check columns separately:

df['baz'] = (pd.to_numeric(df['foo'], errors='coerce').ge(0) | 
            pd.to_numeric(df['bar'], errors='coerce').ge(0)).astype(int)

Thanks, Zero for solution for check numeric:

df['baz'] = df.apply(pd.to_numeric, errors='force').notnull().all(1).astype(int)

---

But if numeric with strings is necessary compare type:

df = pd.DataFrame({'foo': ['a', 1, 2, 3], 'bar': ['b', 10, 25, 9]}) 


df['baz'] = (df.applymap(type) == str).all(1).astype(int)
print (df)
  bar foo  baz
0   b   a    1
1  10   1    0
2  25   2    0
3   9   3    0

Detail:

print (df.applymap(type))
             bar            foo
0  <class 'str'>  <class 'str'>
1  <class 'int'>  <class 'int'>
2  <class 'int'>  <class 'int'>
3  <class 'int'>  <class 'int'>