Fit honest intervention policy tree

What it's for

Learns who an intervention actually helps as a rule you can read, using separate samples for structure, effect and evaluation so the rule isn't graded on its own training data.

Extends causal forests and cohort lift into an inspectable rule that says where an intervention works and evaluates that rule out of sample.

What you give it

Inputs split into evidence read from your connected systems, calibration your team owns, and numerical controls that affect precision but never the result's meaning.

{
  "bootstrap_draws": 100,
  "feature_names": [
    "operating_load"
  ],
  "max_depth": 1,
  "min_leaf_per_arm": 4,
  "observations": [
    {
      "features": {
        "operating_load": -1
      },
      "id": "team-week-0",
      "outcome": -1,
      "propensity": 0.5,
      "treated": true
    },
    {
      "features": {
        "operating_load": -1
      },
      "id": "team-week-1",
      "outcome": 0.05,
      "propensity": 0.5,
      "treated": false
    },
    {
      "features": {
        "operating_load": -1
      },
      "id": "team-week-2",
      "outcome": -0.9,
      "propensity": 0.5,
      "treated": true
    },
    {
      "features": {
        "operating_load": -1
      },
      "id": "team-week-3",
      "outcome": 0.15000000000000002,
      "propensity": 0.5,
      "treated": false
    },

What you get back

This is the actual output of running the example above — computed by the same function the platform calls, not an illustration.

{
  "assumptions": [
    "Outcomes are oriented so larger values are better.",
    "Propensities are known or consistently estimated and all confounders are observed.",
    "Units do not interfere with each other and treatment definitions are stable.",
    "Rules apply to teams or cohorts and must not automate individual employment decisions."
  ],
  "feature_importance": [
    {
      "feature": "operating_load",
      "share": 1
    }
  ],
  "leaves": [
    {
      "control_count": 10,
      "effect": -0.95,
      "effect_interval_90": {
        "high": -0.8917,
        "low": -1.0002
      },
      "leaf_id": "leaf-0",
      "probability_effect_exceeds_minimum": 0,
      "recommend_treatment": false,
      "status": "estimated",
      "treated_count": 6
    },
    {
      "control_count": 8,
      "effect": 2.0208,
      "effect_interval_90": {
        "high": 2.0627,
        "low": 1.9706
      },
      "leaf_id": "leaf-1",
      "probability_effect_exceeds_minimum": 1,
      "recommend_treatment": true,
      "status": "estimated",
      "treated_count": 12
    }
  ],
  "method": "three_way_honest_ipw_intervention_policy_tree_v1",
  "out_of_sample_policy_evaluation": {
    "effective_sample_size": 15,

How it works

Constrained optimization — Pick the best feasible option under real limits — budget, headcount, dependencies, capacity — rather than ranking a list and hoping it fits.

1. 1 Learn an interpretable heterogeneous intervention rule using separate structure, effect-estimation, and untouched policy-evaluation samples.
2. 2 Evaluate the method-specific diagnostics and gates returned by the function, then abstain unless the declared decision clears them under locally governed thresholds.

Before you trust it

Every tool in the catalog ships with the conditions under which its answer is meaningful — and the conditions under which it should abstain instead of guessing.

Calibrating it to your org

Calibration workflow

1. 1 Define the management decision, target outcome, aggregate unit, privacy boundary, cadence, and prediction/intervention horizon for this organization.
2. 2 Build a tenant-scoped historical cohort using only information available before each prediction or decision; preserve zero periods, censoring, assignment probabilities, and unresolved outcomes when the method requires them.
3. 3 Estimate statistical parameters on training history, but obtain costs, utilities, risk tolerance, practical-effect thresholds, capacity, and policy constraints from accountable decision owners.
4. 4 Validate on later time windows or held-out aggregate units at the deployment grain, against a simple baseline and the function-specific validation strategy.
5. 5 Deploy only if the returned decision clears evidence, overlap, calibration, robustness, and guardrail checks; warning, unsupported, schema-gap, and fallback decisions are abstentions.
6. 6 Monitor realized outcomes, data drift, coverage, and decision regret; recalibrate at a governed cadence or after a detected regime/definition change, never merely because a stakeholder dislikes the result.

Call it from your AI

You don't wire up 388 tools in your MCP client. The GitRevio MCP server exposes 18 tools, three of which let an agent search the catalog, read a tool's schema, and run it — so the assistant finds this one on its own.

gitrevio_capabilities_search
  { "q": "learn an interpretable heterogeneous intervention rule" }
  → finds "fit_honest_intervention_policy_tree"

gitrevio_capability_describe
  { "capability_id": "fit_honest_intervention_policy_tree" }
  → returns the input schema and agent guidance shown on this page

gitrevio_capability_run
  { "capability_id": "fit_honest_intervention_policy_tree", "arguments": { ... } }
  → returns the result shown above

Works in Claude Desktop, Claude Code, Cursor, Cline, Continue.dev, Goose and Aider. See the MCP server.

Related tools

See every tool in Causal evidence & experiments →