Segments and Classifications

Segments

Route: /segments

Segments are named groups of item-site series sharing common characteristics. A segment is defined using a recursive criteria builder and can be used to:

The Segments page lists all configured segments and the number of matching series.

• Filter the Dashboard, Series table, and Exceptions screens
• Assign different MEIO parameters per group
• Apply different forecasting parameter sets per group
• Scope exception filtering
• Define ABC/XYZ cross-combinations (e.g. AX parts)

---

Creating a Segment

1. Navigate to Segments in the sidebar.
2. Click + New Segment.
3. Enter a name and optional description.
4. Build criteria using the criteria builder.
5. Click Preview to count matching series.
6. Click Save.

The Criteria Builder

Each condition has three parts: Field, Operator, Value. Conditions combine with AND or OR. Groups can be nested.

Available field categories:

Category	Examples
Item	Item Code, Item Name, Item Type, custom item attributes
Site	Site Code, Site Name, Site Type, custom site attributes
Demand / Series	is_intermittent, has_seasonality, has_trend, complexity_level, n_observations, mean
Classification	One entry per active ABC/XYZ config (e.g. Default (demand), XYZ (order hits))

The Classification group appears in the field dropdown when at least one ABC/XYZ configuration has been run. Each classification field offers the class labels defined in that config (e.g. A/B/C or X/Y/Z) as enum options.

Example: AX parts

AND
  classification.Default (demand)  =  A
  classification.XYZ (order hits)  =  X

This selects all series classified A in the demand-volume scheme AND X in the order-frequency scheme.

Order frequency parameter overrides

Segments can also be used to assign different order_frequency parameter sets to different item groups. For example, high-volume items might have weekly + monthly ordering, while slow movers are limited to quarterly + annually. This is configured in Settings → Parameters → Order Frequency by assigning a parameter set to one or more segments. See Order Frequency Parameters.

Pre-built combination segments are provided out-of-the-box:

Segment	Criteria
AX parts	Default (demand) = A AND XYZ (order hits) = X
BX parts	Default (demand) = B AND XYZ (order hits) = X
AY parts	Default (demand) = A AND XYZ (order hits) = Y
CZ parts	Default (demand) = C AND XYZ (order hits) = Z

Classification field names

The field key is classification.{config_name}. If a classification config is renamed, segments using the old name must be updated manually.

---

ABC / XYZ Classification

Route: /pipelines → Classification tab (leftmost tab)

ABC and XYZ are independent classification schemes. Multiple schemes can run simultaneously. Each scheme produces a class label per item-site combination that feeds into:

• Segment criteria (filter by class combination)
• Dashboard series table (one column per scheme)
• Combined results table (see below)

Default Configurations

Config	Metric	Labels	Meaning
Default (demand)	Demand volume	A / B / C	A = top 80% of cumulative demand
XYZ (order hits)	Order hits	X / Y / Z	X = top 80% by order count

Combined Results Table

The /abc page (Classifications) shows a combined table — one row per SKU, one column per active classification. Use the filter chips in the header to filter by class across multiple schemes simultaneously:

• Click A on "Default (demand)" → show only A-class items
• Click X on "XYZ (order hits)" → combine: show AX items only
• Filters are AND-combined across schemes

Running Classifications

From /pipelines → Classification tab:

1. Select a configuration card on the left.
2. Click Run to compute classes for the active pipeline.
3. Click Run All to run all active configurations at once.

After running, results appear in: - The combined table on /abc - The Dashboard series table (one column per config) - As criteria options in the Segment builder

The Classifications page shows the combined ABC/XYZ results table with filter chips.

Configuring a Classification

Click + New Classification to create a new scheme:

Field	Description
Name	Display name (also used as the classification.{name} field key in segment criteria)
Metric	hits (order count), demand (qty sum), value (qty × price)
Lookback	Months of history to use
Granularity	Per item/site or per item (all sites aggregated)
Method	cumulative_pct (Pareto), rank_pct (top N%), rank_absolute
Class labels & thresholds	Define how many classes and where the cut-offs are
Dynamic anchor	If enabled, each series uses its own last N months rather than a fixed global window

---

Items Can Belong to Multiple Segments

A series can match multiple segments simultaneously. When multiple segments conflict on a parameter (e.g. MEIO fill rate), the system uses the priority order configured in MEIO Scenarios — higher rows win.