Numpy where in function

October 6, 2022

Hi dear I’m in trouble using a numpy.where function to modify certain values in a pandas dataframe. I want to be clear: if I run this function in a single notebook cell, it runs well, but if i insert it into a function doesn’t.
The condition I try to implement is: if a certain row in a certain column has a certain value, so change the value na in the same row, but in other column, otherwise let’s unchanged the value.
If i write:

df.speed_limit = np.where(df.way.str.contains('link'), df.speed_limit.fillna(40), df.speed_limit)

it runs and does what i would, but if i write:

def change_speed_values(df):
    df.speed_limit = np.where(df.way.str.contains('link'), df.speed_limit.fillna(40), df.speed_limit)
    df.speed_limit = np.where(df.way.str.contains('track'), df.speed_limit.fillna(50), df.speed_limit)
   return df

it runs, but really didn’t do any changes. So you could help me to understand why it happens?

Thank you for your patience and your support. I wish you a good day!

>Solution :

I test it and working well, added alternative solution with numpy.select:

df = pd.DataFrame({'way':['link1','link2','track1','track2'],
                   'speed_limit':[np.nan, 2] * 2})
print (df)
      way  speed_limit
0   link1          NaN
1   link2          2.0
2  track1          NaN
3  track2          2.0

df.speed_limit = np.select([df.way.str.contains('link'),
                            df.way.str.contains('track')],
                           [df.speed_limit.fillna(40),
                            df.speed_limit.fillna(50)], df.speed_limit)

print (df)
      way  speed_limit
0   link1         40.0
1   link2          2.0
2  track1         50.0
3  track2          2.0

df = pd.DataFrame({'way':['link1','link2','track1','track2'],
                   'speed_limit':[np.nan, 2] * 2})
# print (df)

def change_speed_values(df):
    df.speed_limit = np.where(df.way.str.contains('link'), df.speed_limit.fillna(40), df.speed_limit)
    df.speed_limit = np.where(df.way.str.contains('track'), df.speed_limit.fillna(50), df.speed_limit)
    return df

df = change_speed_values(df)
print(df)
      way  speed_limit
0   link1         40.0
1   link2          2.0
2  track1         50.0
3  track2          2.0