Overview
The Process Variants filter selects or removes cases based on the frequency of their process variants. A process variant represents a unique sequence of activities that cases follow through the process. This filter enables you to focus on the most common process paths or identify outlier behavior by filtering cases according to variant frequency.
The filter operates at the case level, keeping or removing entire cases based on their variant ranking. You can select top variants by count (e.g., top 5 variants), by case percentage (e.g., variants representing 80% of cases), or by variant percentage (e.g., top 20% of variants). This makes it valuable for process standardization, Pareto analysis, and outlier detection.
Common Uses
- Process Standardization: Focus analysis on the most common process paths to understand and optimize standard workflows.
- Pareto Analysis: Identify the small number of variants that handle the majority of cases to prioritize improvement efforts.
- Outlier Detection: Remove common variants to analyze exceptional cases and understand deviation patterns.
- Process Simplification: Reduce complexity by filtering to top variants, making process maps and analysis more readable.
- Compliance Analysis: Focus on standard variants to validate compliance, then analyze excluded cases for deviations.
- Performance Benchmarking: Compare performance metrics between the most common variants and less frequent paths.
Settings
Filter Mode: Determines how top variants are selected. Three modes are available:
- Variant Count: Select the top N most frequent variants by count (e.g., 3 = top 3 variants)
- Case Percent: Select variants that account for a specified percentage of total cases (e.g., 0.5 = 50% of cases)
- Variant Percent: Select a specified percentage of the most frequent variants (e.g., 0.25 = top 25% of variants)
Threshold Value: The threshold value interpreted based on the Filter Mode:
- For Variant Count: Number of top variants to select (e.g., 3 for top 3 variants)
- For Case Percent: Percentage of cases to include (e.g., 0.5 for 50%)
- For Variant Percent: Percentage of variants to include (e.g., 0.25 for top 25%)
Remove Filter: When unchecked (default), selects cases matching the criteria. When checked, removes cases matching the criteria, returning only cases that don't match. This allows outlier analysis by excluding the most common variants.
Examples
Example 1: Focus on Top 3 Most Common Variants
Scenario: Your process has 50 different variants, but you want to focus on just the three most common paths to simplify analysis and identify the standard process flow.
Settings:
- Filter Mode: Variant Count
- Threshold Value: 3
- Remove Filter: Unchecked
Result: The filter keeps only cases from the top 3 most frequent process variants, removing all cases that follow other variants.
Insights: This dramatically simplifies process maps and calculators by focusing on the paths most cases follow. If these 3 variants represent 70% of cases, you can optimize these standard paths first for maximum impact.
Example 2: Analyze Variants Representing 80% of Cases
Scenario: You want to apply the Pareto principle (80/20 rule) to your process by focusing on the variants that handle 80% of cases.
Settings:
- Filter Mode: Case Percent
- Threshold Value: 0.8
- Remove Filter: Unchecked
Result: The filter accumulates variants from most to least frequent until reaching 80% of total cases, keeping all cases from those variants.
Insights: This typically reveals that a small percentage of variants (perhaps 10-15%) handle the majority of cases. You can focus process improvements on these high-volume paths while treating less frequent variants as exceptions.
Example 3: Select Top 25% of Variants by Frequency
Scenario: You have 100 different process variants and want to analyze the top quartile to understand common process patterns.
Settings:
- Filter Mode: Variant Percent
- Threshold Value: 0.25
- Remove Filter: Unchecked
Result: The filter selects the top 25 most frequent variants (25% of 100), keeping all cases that follow these variants.
Insights: This provides a middle ground between selecting by case percentage and variant count, useful when you want to analyze a proportional sample of variant diversity.
Example 4: Identify Outlier Cases
Scenario: You want to find exceptional or unusual cases by excluding the two most common process variants.
Settings:
- Filter Mode: Variant Count
- Threshold Value: 2
- Remove Filter: Checked
Result: The filter removes cases from the top 2 most frequent variants, keeping only cases that follow less common paths.
Insights: This helps identify unusual process behavior, potential workarounds, exceptions, or edge cases that may require special handling or indicate process problems.
Example 5: Focus on the "Happy Path"
Scenario: You want to analyze only cases following the single most common process variant to establish a baseline for the ideal process flow.
Settings:
- Filter Mode: Variant Count
- Threshold Value: 1
- Remove Filter: Unchecked
Result: The filter keeps only cases following the most frequent process variant, removing all other cases.
Insights: This reveals the "happy path" - the most common way cases flow through the process. You can use this to establish performance benchmarks and understand what the standard process looks like without deviations.
Example 6: Analyze the Long Tail
Scenario: You want to examine rare variants by excluding the common ones that account for 90% of cases.
Settings:
- Filter Mode: Case Percent
- Threshold Value: 0.9
- Remove Filter: Checked
Result: The filter removes variants representing 90% of cases, keeping only the remaining 10% from rare variants.
Insights: This reveals the "long tail" of rare process variations. These cases may represent errors, special handling, or edge cases that need investigation. Understanding why these variants exist can help prevent them or improve exception handling.
Output
The filter returns a dataset containing cases that match the selected criteria. All cases are complete with their full activity sequences and attributes - the filter operates at the case level and does not remove individual events.
The filtered dataset can be used with calculators to analyze variant-specific performance, duration, or other metrics. When Remove Filter is checked, the output contains the complement set, showing cases that don't match the selection criteria.
Technical Notes
- Filter Type: Case-level filter (removes entire cases, not individual events)
- Variant Definition: A variant is the unique sequence of activities in a case, regardless of timing or other attributes
- Frequency Ordering: Variants are ordered by case count from most to least frequent
- Threshold Handling: For Variant Count mode, if the threshold exceeds available variants, it's capped to the total variant count
- Empty Results: If no variants match the criteria, the filter returns an empty dataset
This documentation is part of the mindzieStudio process mining platform.