to_numeric() returns float values when errors = 'coerce', even if downcast = 'integer' when the list contains 'nan' values

Question

I have the following code,

import pandas as pd
a = ['...', 1]
pd.to_numeric(a, errors = 'coerce', downcast = 'integer')

I expected the output to be,

array([nan,1], dtype = int8)

Instead, I got

array([nan,1.], dtype = float32)

What should I do to get a list of integers? Why does it return float values?

Does this help: https://stackoverflow.com/questions/48530210/pandas-to-numeric-downcast-signed-returning-float64 ? — Mayank Porwal, Apr 18 '20 at 14:52

score 0 · Answer 1 · answered Apr 18 '20 at 15:04

0

type(np.nan) #return float

it is considered as float so the problem in nan value

answered Apr 18 '20 at 15:04

Hamied Amr

Thank You! How could I resolve the issue? As I need Integer values. – Gagan G Apr 18 '20 at 15:08
is (...) value mandatory in your array? your array must be array of numbers – Hamied Amr Apr 18 '20 at 15:48
I am actually using this method on a data source which contains such non-numerical values. – Gagan G Apr 18 '20 at 16:02
you can replace all (...) value with 0 before calling to_numeric so that to_numeric cast it as int8 – Hamied Amr Apr 18 '20 at 16:09
Could you please provide some detail on how to resolve this? This right now is not very helpful as an answer. – creyD Apr 18 '20 at 16:50
``` python import numpy as np arr = np.array(['...','1']) mask = [not(i.isnumeric()) for i in arr] arr[mask] = '0' print(arr) ``` – Hamied Amr Apr 18 '20 at 19:07

1 Answers1