Doubt about converting str to int with py/pandas ((dataset))

Question

I have a column of notes in string format. They come as follows: 432432.0 Always comes with . 0 then I wanted to go straight to int and not to float As ta in this format I saw that I should first convert to float and then to int Only that there are some lines that had the string "Unknown" that I previously changed everything to None Then when I turn the conversation from float to int of this error:

ValueError: Cannot convert non-finite values (NA or inf) to integer

I can solve this error with this modification here:

thereof:

data['Score-9'] = data['Score-9'].astype(int)

for that reason:

data['Score-9'] = data['Score-9'].fillna(0).astype(int)

but I don’t want None’s gaps to be 0 because that’s gonna change the averages, medians and the like. How can I do the conversion by continuing with None and without error?

Answer 1

In average, median and fashion calculation; the value NaN is ignored.

Give preference to the type float.

Take the example:

Creating Dataframe Test

df = pd.DataFrame({"valores": ["1", "1", "Unknown", "4"]})

df
   valores
0        1
1        1
2  Unknown
3        4

Converting Unkown for NaN

import numpy as np

df = df.replace("Unknown", np.nan)

In case you try to convert to int, the exception will be made

df["valores"] = df["valores"].astype(int)
(...)
ValueError: cannot convert float NaN to integer

Converting to float

df["valores"] = df["valores"].astype(float)

df
   valores
0      1.0
1      1.0
2      NaN
3      4.0

Medium, medium, fashion

df["valores"].mean()
2.0

df["valores"].median()
1.0

df["valores"].mode()
0    1.0
dtype: float64

Answer 2

You can use lambda this way to remove ". 0" since your numbers are integers

data['Score-9'] = data['Score-9'].apply(lambda x: x.replace('.0', ''))

Answer 3

What you can do is treat the values "Unknown" to 0 so when convert to int you will not get this error.

If you don’t want the values like 0, after the conversion, change the 0 to None