> For the complete documentation index, see [llms.txt](https://doc.verteego.com/verteego-doc/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.verteego.com/verteego-doc/pipelines/optimization-pipelines/constraints/aggregation-constraint.md).

# Aggregation Constraint

### Overview

Aggregation Constraints in Verteego allow for the enforcement of conditions on aggregated values of variables across specified groups within your data. These constraints are vital for managing totals, averages, or other statistical measures across various dimensions of a dataset, ensuring they adhere to specified business rules or operational limits.

### Applications

* **Vehicle Inventory Management:** The sum of vehicles in each category must not exceed the total available from suppliers.
* **Rental Operations:** Total rental days per month must stay within a predefined range of minimum and maximum days.

### Parameters

* **keys:** Optional. A list of column names that define the grouping for aggregation. If omitted, all selected rows are aggregated into a single value.
  * Example: Using `keys=["car_model", "fuel_type"]` aggregates values separately for each combination of car model and fuel type.
* **where:** Optional. A dictionary of key-value pairs to filter rows before applying the constraint.
  * Example: `where={"month": [1, 2, 3], "year": 2022}` limits aggregation to the first quarter of 2022.
* **left\_method:** Specifies the aggregation method for the left side (e.g., `mean`, `min`, `max`, `np.sum`). See the [Pandas DataFrame aggregation documentation](https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.agg.html) for more methods.
* **left\_column:** Names of one or more columns to aggregate. Multiple columns result in the sum of aggregated values being constrained.
* **left\_column\_weight:** Optional. Lists of weights to apply to each column that gets aggregated.
* **left\_where:** Optional. Filters rows for aggregation on the left side only.
  * Example: `left_where={"bicycle_type": ["push_bike", "tandem"]}` aggregates only rows concerning push bikes and tandem bikes.
* **operator:** Specifies the type of constraint (e.g., `equal`, `lesser`, `greater`, `between`) between the aggregated results on the left and right sides.
* **right\_method, right\_column, right\_column\_weight, right\_where:** These parameters mirror those on the left side but apply to the right side of the equation.
* **relax:** Indicates whether this constraint can be relaxed with a penalty. Defaults to `false`.
* **keep\_duplicates:** Determines whether to keep multiple rows that might refer to the same variables. Defaults to `false`.

### Examples

* **Comparing Rental Prices for Different Bicycle Types**<br>

  ```yaml
  yamlCopy codeconstraint_electric_versus_manual_bicycles:
    constraint_type: aggregation
    keys:
    - bicycle_type
    left_column: rental_price
    left_method: max
    left_where:
      bicycle_type:
        - "push_bike"
    operator: lesser
    right_column: rental_price
    right_column_weight:
      - 0.9
    right_method: min
    right_where:
      bicycle_type:
        - "e_bike"
  ```

* **Checking Hotel Occupancy Limits**<br>

  ```yaml
  yamlCopy codehotel_occupancy_checker:
    constraint_type: aggregation
    keys:
    - hotel_category
    left_column: hotel_occupancy
    left_method: mean
    operator: between
    right_column:
      - min_avg_occupancy_per_category
      - max_avg_occupancy_per_category
    right_method: min
  ```