In Pandas, how do I apply a function to a row of a dataframe, where each item in the row should be passed to the function as an argument?

Question

In other words, say I have a dataframe with some columns, and numerical data in the table. For example, I have height, weight, age. Simple dataframe packed with numbers.

What I want is to make a new series (and add it to the dataframe) that is the result of some calculation using each item from each row. So I have a function f(height, weight, age) and I want the numerical result from that function stored as it's own new column.

So on a given row, I'd have the height, weight, age, and the result of f().

I'm sorry, I explored lots of pandas apply examples and can't find anything that quite does what I have in mind here, though it seems like it's something that should be doable!

Thanks in advance!

Answer 1

Let us take an examle, in which we have a dataframe in which we have weight and height.

We can use the apply function to apply a function on each row with all column or selected columns as follows:

df = pd.DataFrame({"height": [180, 178, 190, 166], 'weight': [78, 72, 89, 75] })

print(df)

   height  weight
0     180      78
1     178      72
2     190      89
3     166      75

def bmi(x):
    return x.weight/((x.height/100)**2)

df['bmi'] = df.apply(lambda x: bmi(x), axis=1)

print(df)

   height  weight        bmi
0     180      78  24.074074
1     178      72  22.724403
2     190      89  24.653740
3     166      75  27.217303

Answer 2

Now if i understood correctly, your function calculated a value using your data (height, weight, and age) for each row and that will be in a new column in the dataframe, am i right?

Now you want to do that row by row, which i'm not sure why? do you want to iteritate over your dataframe? I don't have enough information about your function to tell whether that's really needed or not, but generally i'd avoid that approach as it's much slower than running a victorisation process like below:

df['result'] = #whatever your function is doing using the df['height', 'weight', 'age']

example, let's assume your function multiplies height by weight and then divides by age, so you can do the following:

df['result'] = (df['height'] * df['weight']) / df['age']