为什么 pd.read_csv('file.csv') add.999999 在某些值的末尾？

Question

I have a csv file with 4 rows and 1 column. When I open it with Sublime it looks like this:

2.291433301000000000e+09
3.601532401000000000e+09
3.061400502000000000e+09
3.195901470100000000e+10

When I read it using:

df = pd.read_csv('file.csv', names=['Column 1'])

The value at the last row in python is 31959014700.999996

How can I solve the issue? I tried adding data type when reading the file:

df = pd.read_csv('file.csv', names=['Column 1'], dtype=np.int64)

But didn't work. I also tried:

df = pd.read_csv('file.csv', names=['Column 1'])
df = df(pd.to_numeric, errors='coerce')

But it says it cannot convert to int!

Thanks for your help.

Answer 1

That is a floating point error, and Pandas is refusing to convert because it won't round floats automatically.

Try this:

df = pd.read_csv(
    'file.csv', names=['Column 1']
).round(0).astype(int)

If you are writing data into the CSV as well and only plan to store integers, you might not want to use scientific notation. Numbers in scientific notation will be interpreted as floating point, so you will have to do this to represent them as integers without occasional failure.