Perform a “fuzzy” join between the 2 input tables.

Parameters:
A typical Mobile Money transaction goes, basically, the following way:
We want to find, for each transaction of TABLE A, the ANTENNA_ID from the subscriber that started the transaction. In other word, we need to compute a join between the TABLE A and the TABLE B.
We could try to compute the join using the standard Join action. More precisely:
This approach is described in more detail in section 5.1.6 about Composite primary Keys. This naïve approach will fail because the “TIME” column is not exactly the same inside each of the two tables (because these are two different servers that are generating the two tables A and B and their clock is not tightly synchronized).
To be able to compute the join, we need a “Fuzzy” match on the TIME dimension. Meaning that, if the TRANSATION_TIME and the TIME are separated by less than 3 minutes (this is the “Time Window” parameter), we’ll decide that the TIME columns are “matching” (and we’ll thus produce a join and retrieve the ANTENNA_ID associated with the transaction).
The parameters of the JoinCDR action are:
The JoinCDR action is written in JavaScript so you can easily tweak/modify it to suit your specific needs.
