AbcClassification

DaxPatterns.AbcClassification

Functions to implement the ABC Classification pattern (dynamic version) from DAX Patterns.

What it does

ComputeInAbcClass evaluates an expression by filtering the entity (e.g., Product, Customer) belongs to the ABC classes active in the filter context from a disconnected “ABC Classes” table, respecting all active report filters. Use it to build dynamic ABC analyses that slice by time, region, or any other dimension.

---

Function

DaxPatterns.AbcClassification.ComputeInAbcClass (
    <valueExpr>,               -- e.g., [Sales Amount] (must be additive, SUM-based)
    <transactionsTable>,       -- e.g., Sales
    <itemTable>,               -- e.g., 'Product'
    <itemKeyColumn>,           -- e.g., 'Product'[ProductKey]
    <abcClassTable>,           -- e.g., 'ABC Classes' (disconnected)
    <abcLowerBoundaryColumn>,  -- e.g., 'ABC Classes'[Lower Boundary]  -- inclusive
    <abcUpperBoundaryColumn>   -- e.g., 'ABC Classes'[Upper Boundary]  -- exclusive (see notes)
)

Return: TRUE if the entity’s cumulative share (by <valueExpr>) falls within the class boundaries on the current row of <abcClassTable>. Otherwise FALSE. Use it inside measures to filter or label entities.

Assumptions: Dynamic ABC as in the DAX Patterns article with a parameter table of class boundaries. The measure supplied must aggregate with SUM.

---

Model requirements

1. Value measure: Additive, e.g.:

   Sales Amount = 
   SUMX ( Sales, Sales[Quantity] * Sales[Net Price] )
   

2. Entity table: The table that lists the items to classify (e.g., Product).

3. Classes table (disconnected): No relationships. Contains at least:

   • Class (A/B/C or any label)
   • Lower Boundary (0–1)
   • Upper Boundary (0–1)

Example ABC Classes (typical Pareto configuration):

ABC Classes = 
DATATABLE (
    "Class", STRING,
    "Lower Boundary", DOUBLE,
    "Upper Boundary", DOUBLE,
{
    { "A", 0.00, 0.70 },
    { "B", 0.70, 0.90 },
    { "C", 0.90, 1.00 }
})

This matches the pattern described in DAX Patterns and its dynamic variant (classification driven by a disconnected parameter table).

---

Basic usage

ABC Sales Amount =
DaxPatterns.AbcClassification.ComputeInAbcClass (
    [Sales Amount],
    Sales,
    'Product',
    'Product'[ProductKey],
    'ABC Classes',
    'ABC Classes'[Lower Boundary], 
    'ABC Classes'[Upper Boundary]
)

These produce a dynamic ABC where class membership updates with filters (e.g., year, region).

---

Tips, constraints, and performance

• Additivity: <valueExpr> must be SUM-based for correct cumulative ranking in the dynamic pattern. Non-additive measures break the logic.
• Boundaries: Treat Lower Boundary as inclusive and Upper Boundary as exclusive to avoid overlaps at exact cut points (e.g., 0.70 belongs to B if A ends at 0.70).
• Disconnected classes: Do not relate ABC Classes to other tables. It is a parameter table used by the measure.
• Granularity: Pass the entity key that defines classification granularity.
• Scaling: With very high entity cardinality, consider pre-aggregation by entity or snapshot ABC when dynamics are not needed.

---

Related reading

• DAX Patterns: ABC classification (static, snapshot, dynamic).
• DAX Patterns: Dynamic ABC Classification (concepts and step-by-step).
• DAX Patterns: Parameter Table pattern.
• DAX Lib package list (package repository).

---

License

MIT License.