Pandas - convert header name in tuple format to a string

Question

After computing percentiles within group, the header names is in tuple format like [('A', 0.5), ('A',0.9)...('Z',0.9)].

The desired output should be:

['P50 A', 'P90 A', ...'P90 Z']

Basically, I want to multiply the decimal by 100 to get percentage and move it up front, append a letter 'P' in every field.

I feel like I should use map or join like suggested here: How to change the columns name from a tuple to string?

But not sure how to deal with the details.

Can you show how you came up with these in the first place? Looks like you did something "strange" here. — willeM_ Van Onsem
– willeM_ Van Onsem, Commented Oct 16, 2018 at 18:35
df = df[value + grouper].groupby(grouper).quantile(nth_percentile); df = df.unstack(); df = pd.DataFrame(df.to_records()) — Daniel
– Daniel, Commented Oct 17, 2018 at 1:41

screenpaver · Accepted Answer · 2018-10-16 18:46:32Z

2

This works also:

original_names = [('A', 0.5), ('A',0.9),('Z',0.9)]
new_names = ['P'+str(int(100*y)) + ' ' + x for x,y in original_names]

Result: ['P50 A', 'P90 A', 'P90 Z']

answered Oct 16, 2018 at 18:39

screenpaver

1,1508 silver badges14 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Or with an f-string: new_names = [f'P{int(100*y)} {x}' for x,y in original_names]

quest · Accepted Answer · 2018-10-16 18:37:06Z

1

Current column names:

col_name = [('A', 0.5), ('A',0.9),('Z',0.9)]

Desired column names:

desired_col_named = [f'P{x[1]*100} {x[0]}' for x in col_name]

Assuming you have python 3.6 supporting f strings.

answered Oct 16, 2018 at 18:37

quest

3,9762 gold badges18 silver badges27 bronze badges