The transactional analysis wizard currently available in FastStats can be used to find the most frequently occurring patterns of the same length. The limitations in this technique are that it cannot find specific patterns, and cannot find those patterns if they are not the most frequently occurring or if they need to have more specific advanced features (such as wildcards).
The existing Relative aggregations can be used to take information from one transactional record and push it onto another record. In this way, expressions with several ‘AND’ conditions cyprus mobile number can be used together to create a pattern. This technique is very powerful as it can find specific patterns, and ones with some of the more advanced wildcards. ‘*’ wildcards in them, or date restrictions across patterns.
Pattern matching
The pattern matching technique introduced here allows the specification of a set of patterns of potentially different lengths, which are defined in a priority order. A pattern is defined as a set of values that represent the transactions in the sequence. The name associated with the first matched one is returned. This aggregation could be used on its own, or combined together with any of the other expression functionality for more sophisticated analysis.
The option is available in the function type dropdown for an on-the-fly aggregation as ‘Pattern Match’ in FastStats. The user chooses which variable is used to sequence the transactions and the variable to be used for the pattern matching. An example of a set of patterns is shown below.