列出的熊猫数据框系列-抑制浮动科学计数法

Question

I have a Pandas DataFrame with a float column that I convert to a list, then convert to a string, and then write to a text file for another use.

For example:

df=pd.DataFrame([[0.0068149439999999999, 0.90550613999999996], [7.5699999999999997e-05, 0.48159182100000003], 
          [0.009679478, 0.50158892700000002], [0.020792602, 0.56266469899999993]], columns=['ColumnA', 'ColumnB'])

TextToWrite='ColumnA = (' + str(df['ColumnA'].tolist())[1:-1] + ')'
f=open('myfile.txt', 'w')
f.write(TextToWrite)

However, I need the converted floats to not be in Scientific notation (7.5699999999999997e-05, in this example). What is the best way suppress scientific notation for these floats and does it make more sense to do it upfront in the Pandas DataFrame or after the series is already converted to a list?

I looked into the "float_format" argument, which can be done with "to_csv" and "to_string" commands, but I'm not sure how I can do that and still write the string to the text file in the same format I do above.

Answer 1

Ordinarily you can use astype(str) , which is better for rounding. But since your numbers are so small, you'll have to explicitly suppress scientific notation via '%f' .

def to_str(x): return '%f' % x
', '.join(df.ColumnA.apply(to_str).values)

Answer 2

(just realized I'm late with an answer but I'll leave this an an alternate answer if you want finer control over the output format) Just replace the TextToWrite line with:

TextToWrite = 'ColumnA = ' + ' %12.7f'*4 % tuple( df['ColumnA'].tolist() )

to get this:

ColumnA = 0.0068149 0.0000757 0.0096795 0.0207926

Word of caution, you need to choose your format to provide enough precision since this will force decimal formatting over exponential even if the result is 0.0. For example, if you use %12.1 you will just get a bunch of zeroes.

Btw, I believe (from testing) that chrisaycock's %f is equivalent to %8.6f which should be fine here but if you have other values in your data that are smaller (say, .00000009) then you might not get enough precision with %f.