apply using a function with iloc pandas

I have a data set with 40 columns and I want to define a function that operate the columns.

For example,

p = {'val1': [10, 20, 30, 40],
    'val2': [15, 25, 35, 45]}

data = pd.DataFrame(p, columns=['val1', 'val2'])
data

I have this data and I did the following operation

inc = 100*((data.iloc[:, -1]/ data.iloc[:, -1-1])-1)
inc

the result is

0    50.000000
1    25.000000
2    16.666667
3    12.500000
dtype: float64

I want select the max and the index of the max value, I did the following

(inc.idxmax(), max(inc))

I obtained the following result

(0, 50.0)

Now, I define a function

def increase(column):
    inc = 100*((data.iloc[:, -column]/ data.iloc[:, -column-1])-1)
    return (inc.idxmax(), max(inc))

I select the columns backwards.

and I want to apply this function to all my columns

new_data = data.apply(increase)

when I use this I got the error

IndexError: positional indexers are out-of-bounds

and if I use applymap I got the same error

What can I do?

Solution

If I understand right, you want to use multiple columns in the function. The apply is not the solution here because it's working on one row or column (axis=0 or 1) at the time.

so my suggestion is to us for and enter 2 columns per iteration that way:

def increase(col1,col2):
   inc = 100*((col2/ col1)-1)
   return (inc.idxmax(), max(inc))

lst = []
for i in range(len(data.columns)):
    j=i+1
    if j<len(data.columns):
        col1 = data[data.columns[i]]
        col2 = data[data.columns[j]]
        lst.append(increase(col1,col2))

pd.DataFrame(lst)