Overview
The Selected Activity Frequency filter identifies cases based on how many times a specific activity occurs within each case. This case-level filter counts activity occurrences and compares them against a threshold using various comparison operators (greater than, less than, equal to, etc.). The filter is invaluable for detecting rework patterns, repeated activities, quality issues, and process inefficiencies where activities occur more or less frequently than expected.
Unlike filters that check for activity presence or absence, this filter focuses on frequency patterns, allowing you to find cases with excessive repetition or insufficient activity occurrences. It supports both inclusion mode (keep cases matching the criteria) and exclusion mode (remove cases matching the criteria).
Common Uses
- Rework Detection: Identify cases where approval or review activities were repeated multiple times, indicating rework or process inefficiencies.
- Quality Analysis: Find cases with multiple quality check activities, suggesting potential quality issues or excessive scrutiny.
- Exception Handling: Locate cases with frequent error-handling activities, highlighting problematic cases requiring attention.
- Process Compliance: Ensure certain activities occur exactly once or within acceptable frequency ranges for regulatory compliance.
- Anomaly Detection: Discover cases with unusual activity repetition patterns that deviate from normal process flow.
- Efficiency Optimization: Identify cases requiring multiple iterations of the same task, revealing bottlenecks or process improvement opportunities.
Settings
Activity: Select the activity name you want to count occurrences of within each case. The dropdown menu shows available activity names with their frequency statistics.
Comparison Operator: Choose how to compare the activity count against the threshold:
- Greater than: Find cases where the activity occurs more than N times
- Less than: Find cases where the activity occurs fewer than N times
- Equal to: Find cases where the activity occurs exactly N times
- Greater than or equal to: Find cases where the activity occurs N or more times
- Less than or equal to: Find cases where the activity occurs N or fewer times
- Not equal to: Find cases where the activity occurs any number of times except N
Count Threshold: Enter the number to compare against (e.g., 1, 2, 3).
Keep/Remove: Choose whether to keep cases matching the criteria or remove them.
Note: Activity name comparison is case-sensitive and requires exact matching. The filter's validation system will suggest corrections for misspelled activity names.
Examples
Example 1: Finding Cases with Multiple Approvals
Scenario: You want to identify purchase orders where the "Approve Purchase Order" activity occurred more than once, indicating rework or multiple approval rounds.
Settings:
- Activity: "Approve Purchase Order"
- Comparison Operator: Greater than
- Count Threshold: 1
- Keep/Remove: Keep matching cases
Result: The filter returns cases where "Approve Purchase Order" occurred 2 or more times.
Insights: These cases may indicate:
- Initial approval rejections requiring resubmission
- Changes to purchase orders requiring re-approval
- Process inefficiencies in the approval workflow
- Training opportunities for approvers
- Potential cost impacts from delayed approvals
Example 2: Excluding Excessive Resubmissions
Scenario: You want to remove cases where "Submit Application" occurred more than 3 times, as these represent outliers with excessive resubmissions that may distort your analysis.
Settings:
- Activity: "Submit Application"
- Comparison Operator: Greater than
- Count Threshold: 3
- Keep/Remove: Remove matching cases
Result: The filter excludes cases with 4 or more "Submit Application" activities, keeping only cases with 0-3 submissions.
Insights: This helps:
- Focus analysis on typical cases
- Remove extreme outliers
- Identify a separate cohort for special investigation
- Improve average performance metrics
Example 3: Verifying Single Payment Transactions
Scenario: You need to ensure that "Process Payment" occurred exactly once per case, as multiple payments could indicate refunds, corrections, or data quality issues.
Settings:
- Activity: "Process Payment"
- Comparison Operator: Equal to
- Count Threshold: 1
- Keep/Remove: Keep matching cases
Result: The filter returns only cases with exactly one "Process Payment" activity.
Insights: Cases excluded from this filter (with 0 or 2+ payments) should be investigated for:
- Missing payment activities
- Duplicate payment processing
- Refund or correction scenarios
- Data quality problems
Example 4: Identifying Cases Requiring Quality Intervention
Scenario: Find cases where "Quality Check" occurred more than twice, suggesting products with quality issues requiring multiple inspections.
Settings:
- Activity: "Quality Check"
- Comparison Operator: Greater than
- Count Threshold: 2
- Keep/Remove: Keep matching cases
Result: The filter selects cases with 3 or more "Quality Check" activities.
Insights: These cases may reveal:
- Products with recurring quality defects
- Supplier quality issues
- Process capability problems
- Need for root cause analysis
- Training requirements for production staff
Example 5: Finding Cases with At Least One Occurrence
Scenario: Identify all cases where "Send Reminder" occurred at least once, regardless of exact frequency.
Settings:
- Activity: "Send Reminder"
- Comparison Operator: Greater than
- Count Threshold: 0
- Keep/Remove: Keep matching cases
Result: The filter returns cases containing the "Send Reminder" activity one or more times.
Insights: This shows:
- Cases requiring follow-up reminders
- Customer responsiveness patterns
- Communication effectiveness
- Potential process delays
Example 6: Excluding Cases Without Activity
Scenario: Remove cases where "Customer Contacted" never occurred (appears 0 times).
Settings:
- Activity: "Customer Contacted"
- Comparison Operator: Less than or equal to
- Count Threshold: 0
- Keep/Remove: Remove matching cases
Result: The filter removes cases with 0 occurrences of "Customer Contacted", keeping only cases where customer contact happened at least once.
Insights: This ensures your analysis focuses on cases with customer interaction, excluding:
- Automated processing cases
- System-generated cases
- Internal-only workflows
Output
The filter returns a new dataset containing only the cases that match the specified frequency criteria. Each returned case preserves all its original events and attributes. The specified activity may appear 0, 1, or multiple times depending on the filter criteria.
If no cases match the criteria, the filter returns an empty result set.
The output can be further analyzed to understand patterns in activity repetition, such as:
- Average number of repetitions in matching cases
- Time between repeated activity occurrences
- Correlation between activity frequency and other attributes
- Impact of activity repetition on case duration or cost
This documentation is part of the mindzieStudio process mining platform.