Sequential Pattern Miner

Description

Mine bounded frequent ordered subsequences of single-event string sequences using prefix projection, allowing arbitrary gaps and repeated events and counting each source sequence at most once per pattern.

Sequential pattern mining finds event subsequences occurring in enough source sequences. Order matters, but a subsequence can skip events. Sequence-based support counts a source sequence once for a pattern even when that pattern occurs several times within it.

When to use this tool

Use this tool when you need the stated calculation or classification with explicit inputs, intermediate evidence, and boundary conventions.

How to use the tool and how it works

Supply sequences, minimumCount, maximumLength in the field shapes below.
Run the calculation. Correct format, shape, range, or undefined-statistic errors before interpreting a result.
Read sequenceCount, patterns. Use the conventions below.

Grow frequent event prefixes by projected suffixes. For every source sequence, keep the suffix after the earliest occurrence of each possible extension so support counts that sequence at most once. Recurse up to maximumLength and sort patterns by length then serialized lexical event order.^[1]

Input field	Meaning and units	Accepted input
sequences	Sequences	Required. Array of Array of Text; minimum UTF-16 length 1; maximum UTF-16 length 64.; maximum items 32.; minimum items 1; maximum items 128.
minimumCount	Minimum number of source sequences containing a pattern.	Required. Safe integer; minimum 1; maximum 128.
maximumLength	Maximum event count per returned subsequence.	Required. Safe integer; minimum 1; maximum 8.

Output field	Interpretation	Returned shape
sequenceCount	Sequence Count	Required. Safe integer; minimum 1; maximum 128.
patterns	Patterns	Required. Array of Object; no unknown keys.; maximum items 4096.
patterns[].events	Events	Required. Array of Text; minimum UTF-16 length 1; maximum UTF-16 length 64.; minimum items 1; maximum items 8.
patterns[].count	Number of source sequences containing these events in order.	Required. Safe integer; minimum 1; maximum 128.
patterns[].support	count divided by all source sequences, including empty sequences.	Required. Finite number; maximum 1; strictly above 0.

json

{
  "input": {
    "sequences": [
      [
        "A",
        "B",
        "A"
      ],
      [
        "A",
        "B"
      ],
      [
        "B",
        "A"
      ]
    ],
    "minimumCount": 2,
    "maximumLength": 2
  },
  "output": {
    "sequenceCount": 3,
    "patterns": [
      {
        "events": [
          "A"
        ],
        "count": 3,
        "support": 1
      },
      {
        "events": [
          "B"
        ],
        "count": 3,
        "support": 1
      },
      {
        "events": [
          "A",
          "B"
        ],
        "count": 2,
        "support": 0.6666666666666666
      },
      {
        "events": [
          "B",
          "A"
        ],
        "count": 2,
        "support": 0.6666666666666666
      }
    ]
  }
}

Limitations and assumptions

Patterns are ordered subsequences with arbitrary gaps and may repeat an event; they are not necessarily contiguous. Support divides count by all supplied sequences, including empty ones. The alphabet is limited to sixteen events and work/pattern budgets can reject oversized mining.
The field tables state structural bounds; cross-field constraints and bounded search or output budgets also apply. Unknown keys are rejected where the object contract disallows them. Input errors are distinct from null fields and deliberate invalid controls in a successful output.

Alternative or Complementary methods

Choose the statistic according to the question and data conventions. Inspect the supplied series or matrix and the stated assumptions before comparing outputs from implementations with different defaults.

References

Tool behavior and example: Sequential Pattern Mining Tool — tool.io

Similar or alternative tools

FP-Growth Frequent Itemset Mining Tool
Mine bounded transaction itemsets with actual FP-trees and weighted conditional pattern bases, counting each item once per transaction and retaining empty transactions in support denominators.
Apriori
Find item combinations whose support meets a selected threshold.