find_frequent_subsequences()

find_frequent_subsequences() finds event patterns that appear often enough in your event sequences. It is the core mining step of this section: you set a support threshold (a minimum number or share of sequences that must contain a pattern), and the function returns every subsequence that clears it.

What It Does

• Searches the event sequences for subsequences whose support reaches min_support (count) or min_support_ratio (share).
• How support is counted depends on search_constraint; with timestamps, the constraint can also restrict matches by time span or by the gap between events.
• With target_subsequences, it skips discovery and counts only the patterns you name.
• Returns a SubsequenceList that downstream functions such as compare_groups() and the plotting functions consume.

Returns

sequenzo.event_sequences.core.SubsequenceList. SubsequenceList object with frequent subsequences

Function Usage

find_frequent_subsequences(
    event_sequences,
    target_subsequences=None,
    min_support=None,
    min_support_ratio=None,
    search_constraint=None,
    max_k=-1,
    weighted=True
)

TraMineR Parameter Mapping

• event_sequences -> TraMineR eseq object
• target_subsequences -> TraMineR::seqefsub() str.subseq
• min_support_ratio -> TraMineR::seqefsub() pmin.support
• search_constraint -> TraMineR::seqefsub() constraint
• weighted -> TraMineR::seqefsub() weighted

Entry Parameters

Parameter	Required	Type	Description
event_sequences	✓	EventSequenceData / EventSequenceList	Your full event-sequence dataset (the object returned by EventSequenceData.from_tse() or EventSequenceData.from_state_sequences()).
target_subsequences	✗	list[str]	Optional list of specific subsequences to search for directly.
min_support	✗	float	Minimum number of supporting sequences (or weighted support).
min_support_ratio	✗	float	Minimum support ratio (0 to 1).
search_constraint	✗	EventSequenceConstraint	Time and counting constraints.
max_k	✗	int	Maximum number of events in a subsequence (-1 means no limit).
weighted	✗	bool	Use sequence weights if available.

Examples

# At least 20 supporting sequences
fsubseq = find_frequent_subsequences(event_sequences, min_support=20)

# At least 5% support
fsubseq = find_frequent_subsequences(event_sequences, min_support_ratio=0.05)

# Search only selected patterns
fsubseq = find_frequent_subsequences(
    event_sequences,
    target_subsequences=["(EnterUni)-(Graduate)", "(Graduate)-(FindJob)"]
)

R Counterpart

• Closest R function: TraMineR::seqefsub()
• Mapping note: Both functions search for frequent event subsequences using support thresholds and optional time constraints.

Notes

• You must provide min_support or min_support_ratio when target_subsequences is not given.
• A subsequence match preserves event order, but the events do not have to be adjacent unless this is imposed through search_constraint.
• If timestamps are available, search_constraint can restrict matches by time span or time gap.
• Support depends on the counting rule. For example, a presence-based rule counts whether a sequence contains the subsequence, while an occurrence-based rule may count repeated occurrences.
• Weighted support uses sequence weights from event_sequences.

See Also

• Event Sequences Introduction explains the workflow and data structures.
• Quickstart Example shows a complete event-mining run.
• Helper Functions lists supporting utilities.

Authors

Code: Yuqi Liang

Documentation: Yuqi Liang

References

Ritschard, G., Bürgin, R., & Studer, M. (2013). Exploratory Mining of Life Event Histories. In J. J. McArdle & G. Ritschard (Eds.), Contemporary Issues in Exploratory Data Mining in the Behavioral Sciences (pp. 221-253). Routledge.