> ## Documentation Index > Fetch the complete documentation index at: https://docs.woodwide.ai/llms.txt > Use this file to discover all available pages before exploring further. # Estimate Causal Effect > Estimate the average effect of `do(treatment = treatment_action)` vs `do(treatment = control_action)` for a **causal** model, by g-computation over the model's training dataset (the causal frame is rebuilt from the labels frozen at training time). Always asynchronous: returns a `job_id` immediately (202). Poll `GET /jobs/{job_id}` until `succeeded`, then fetch the scalar results (`estimand`, `effect`, `n_rows`, `n_extrapolated`, `tau_summary`) from the `summary` of `GET /jobs/{job_id}/results`; the per-row effects (tau — the CATE surface) are served by `GET /jobs/{job_id}/results/rows`. `estimand`: - `ate`: average treatment effect over every training row. - `att`: average effect on the treated (rows whose observed treatment equals `treatment_action`). Categorical treatments only — ATT for continuous treatments requires a dose band and is not yet supported (422). Unanswerable requests fail with 422 up front: a categorical action the model never saw at training, or a non-numeric action for a continuous treatment. Identical requests return the previously dispatched job. ## OpenAPI ````yaml /openapi.json post /models/{model_id}/estimate-causal-effect openapi: 3.1.0 info: title: Woodwide API description: API service for the Woodwide platform version: 0.1.0 servers: - url: https://api.woodwide.ai description: Production security: [] paths: /models/{model_id}/estimate-causal-effect: post: tags: - Inference summary: Estimate Causal Effect description: >- Estimate the average effect of `do(treatment = treatment_action)` vs `do(treatment = control_action)` for a **causal** model, by g-computation over the model's training dataset (the causal frame is rebuilt from the labels frozen at training time). Always asynchronous: returns a `job_id` immediately (202). Poll `GET /jobs/{job_id}` until `succeeded`, then fetch the scalar results (`estimand`, `effect`, `n_rows`, `n_extrapolated`, `tau_summary`) from the `summary` of `GET /jobs/{job_id}/results`; the per-row effects (tau — the CATE surface) are served by `GET /jobs/{job_id}/results/rows`. `estimand`: - `ate`: average treatment effect over every training row. - `att`: average effect on the treated (rows whose observed treatment equals `treatment_action`). Categorical treatments only — ATT for continuous treatments requires a dose band and is not yet supported (422). Unanswerable requests fail with 422 up front: a categorical action the model never saw at training, or a non-numeric action for a continuous treatment. Identical requests return the previously dispatched job. operationId: estimate_effect_models__model_id__estimate_causal_effect_post parameters: - name: model_id in: path required: true schema: type: string title: Model Id requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/EffectRequest' responses: '202': description: Successful Response content: application/json: schema: $ref: '#/components/schemas/InferBatchResponse' '422': description: Validation Error content: application/json: schema: $ref: '#/components/schemas/HTTPValidationError' security: - HTTPBearer: [] components: schemas: EffectRequest: properties: estimand: $ref: '#/components/schemas/Estimand' description: 'Which effect to estimate: ate or att.' treatment_action: anyOf: - type: number - type: integer - type: boolean - type: string title: Treatment Action description: Counterfactual treatment value for the 'treated' arm. control_action: anyOf: - type: number - type: integer - type: boolean - type: string title: Control Action description: Counterfactual treatment value for the control arm. model_version_id: anyOf: - type: string - type: 'null' title: Model Version Id description: >- Public ID of a specific model version (e.g. mdlv_A8K2P9QX). If omitted, the latest ready version is used. type: object required: - estimand - treatment_action - control_action title: EffectRequest description: |- Request body for causal effect estimation (``POST /models/{id}/estimate-causal-effect``). Estimates the average effect of ``do(treatment = treatment_action)`` vs ``do(treatment = control_action)`` by g-computation over the model's training dataset (the frame is rebuilt from the frozen causal labels). Always asynchronous: returns a job to poll; scalar results land in the job's results ``summary``, per-row effects (tau) via ``GET /jobs/{id}/results/rows``. InferBatchResponse: properties: job_id: type: string title: Job Id description: Public ID of the inference job (e.g. job_A8K2P9QX). status: type: string title: Status model_version_id: type: string title: Model Version Id description: Public ID of the model version used (e.g. mdlv_A8K2P9QX). type: object required: - job_id - status - model_version_id title: InferBatchResponse description: Async inference submission response. HTTPValidationError: properties: detail: items: $ref: '#/components/schemas/ValidationError' type: array title: Detail type: object title: HTTPValidationError Estimand: type: string enum: - ate - att title: Estimand description: |- Supported causal estimands. - ``ate``: average treatment effect over every training row. - ``att``: average effect on the treated (rows whose observed treatment equals ``treatment_action``). Categorical treatments only — ATT for a continuous treatment requires a dose band, which is not yet supported. ValidationError: properties: loc: items: anyOf: - type: string - type: integer type: array title: Location msg: type: string title: Message type: type: string title: Error Type input: title: Input ctx: type: object title: Context type: object required: - loc - msg - type title: ValidationError securitySchemes: HTTPBearer: type: http scheme: bearer ````