NaN and +/- Inf values can arise in floating point calculations. They are rendered when a DataFrame is evaluated.
>>> df value row_id 2 -inf 1 inf 3 NaN >>> df.dtypes row_id str value float
NaN and +/- Inf values are not supported as missing values. Particularly, there is no support to reference these values in the Advanced SQL Engine. Only the NULL value is supported as a missing value, in which case they are usually rendered as None. Floating point columns with NULL values can be rendered as NaN. In this case, NaN is recognized as a missing value.
>>> df[df.value.isna() == True] value row_id 3 None >>> df[df.value.isna() == False] value row_id 2 -inf 1 inf