here is my code
df = pd.DataFrame([110100.0], dtype=np.float32)
df.round(7)
the result is:
110099.992188
what I expect is 110100.0.
How can I make sure that the round operation only affect decimal part and the integer part remained unchanged
for example: input => expected output
1.0 => 1.0
1.12345678 => 1.1234567
You can use numpy.modf for extract floats to fractional and integral parts and round only fractional part, last add together:
df = pd.DataFrame([110100.0, 1.0, 1.12345678], dtype=np.float32, columns=['col'])
a, b = np.modf(df['col'])
print (b + np.round(a, 7))
0 110100.000000
1 1.000000
2 1.123457
Name: col, dtype: float32
Another solution is multiple by number for round, convert to integer and divide:
val = 10**7
print (df['col'].mul(val).astype(np.int64).div(val))
0 110100.000000
1 1.000000
2 1.123457
Name: col, dtype: float64
Use round function of python and give precision EG:
a = 1.112346789
round(a, 6)
1.112347
round(a, 1)
1.1