- Support empty numerical values in ReadCSV. (195) - in case time is pandas timestamp set timezone manually - null in TS column
0.4.0
Not secure
- Prioritize datetime over int when inferring type in ReadCSV - reading from csv None in boolean column - Aggregations emit NaN values where a value cannot be calculated due to lack of data. - In case key does not exist do not return features default values
0.3.8
Not secure
- Add support for multiple key fields - Remove nanoseconds when reading from CSV & writing to parquet