I am currently working with pandas version 2.2.0 and I have the following issue when rounding my DataFrame:
pd.DataFrame({"value": [26586491.0]}).round(15)["value"].iloc[0][2]: 26586490.999999996This is annoying because there might be situations where we want to compare the exact value. Using numpy directly gives similar issues.Any idea what is wrong here ? I guess the problem comes from how pandas stores floats but is there a way to avoid such behavior ?