使用Pandas处理CSV后，如何移除DataFrame中所有元素的.0后缀（保留非整数小数）

Great question! The .0 suffix you're seeing happens because when you fill NaN values with -99, pandas converts any integer columns to float (since NaN is a float type by default). Here are a few tailored solutions depending on your data structure:

Option 1: Convert entire columns to integer type (best if all values in the column are whole numbers) #

If the columns showing .0 are supposed to be integers (no decimal values like 2.7), you can convert them back to integer dtype. This keeps your data in efficient numeric formats:

import pandas as pd

# Your existing code
data = pd.read_csv(r"\test.csv", sep=';')
data = data.fillna(-99)

# Iterate through columns and convert whole-number float columns to int
for col in data.columns:
    # Check if all values in the column are whole numbers
    if (data[col] % 1 == 0).all():
        data[col] = data[col].astype(int)

Option 2: Convert individual whole-number floats to integers (for mixed columns) #

If some columns have a mix of whole numbers (like 202.0) and actual decimals (like 2.7), you can convert only the whole-number floats to integers while leaving decimals as-is. Note this will change the column dtype to object, which is fine for display but may impact performance on large datasets:

def clean_float_values(x):
    # Convert float to int if it's a whole number
    if isinstance(x, float) and x.is_integer():
        return int(x)
    return x

# Apply the function to every element in the DataFrame
data = data.applymap(clean_float_values)

Option 3: Use nullable integer types (prevent float conversion in the first place) #

If you know which columns should be integers (even with missing values), you can use pandas' nullable Int64 dtype when reading the CSV. This avoids converting to float entirely:

# Specify dtype for integer columns (replace 'col1', 'col2' with your actual column names)
data = pd.read_csv(r"\test.csv", sep=';', dtype={'col1': 'Int64', 'col2': 'Int64'})
data = data.fillna(-99)

If you're not sure which columns are supposed to be integers, you can combine this with the check from Option 1 after filling:

data = pd.read_csv(r"\test.csv", sep=';')
data = data.fillna(-99)

for col in data.columns:
    if data[col].dtype == float and (data[col] % 1 == 0).all():
        data[col] = data[col].astype('Int64')

Quick note on display-only fixes #

If you only care about how the data looks when printed (not the underlying dtype), you can set a custom display format—but this won't change the actual data values:

# This will show whole-number floats without .0, but keep decimals as-is
pd.options.display.float_format = lambda x: '{:.0f}'.format(x) if x.is_integer() else '{:.1f}'.format(x)

To refine this further, it would help to know if the columns with .0 are all intended to be integers, or if some have a mix of whole numbers and decimals. But the above options cover both scenarios!

内容的提问来源于stack exchange，提问作者Omar Fulla