Add to SDE theory Missing Values Methods

[emanueledellavalle]

• Drop: drops all the rows that contain at least one missing value. It’s a Filter operator with a space complexity of O(1) and can be distributed.
• Fill Constant: fills all the missing values with a constant value. It's a Map Operator that finds the missing values and modifies them. It has a space complexity of O(1) and can be distributed.
• Fill Backward/Forward: Fill all the missing values with the last/next valid value of that column. The backward operator needs to save the contiguous data containing missing values so it cannot be streaming, O(n). Both need to have the data in order so they cannot be distributed.
• Fill Statistic: fills all the missing values with a statistic value (max, min, mean) calculated on the column. They need to perform two iterations over the dataset so they are not streaming but can be distributed.