- Home
- Skills
- Research & Analysis
- Causal ML Starter: Method Selector
Causal ML Starter: Method Selector
Choose a causal ML design from your estimand, treatment type and data structure. One complete method-selection skill with equations, identification checks and official EconML/DoubleML references.
Free
Works with the AI Agents you already use
Causal ML Starter: Method Selector
Example session with this skill installed
My treatment is binary, the outcome is continuous and the data are observational. I want the population ATE. Use the method selector to propose an estimand, assumptions, estimator and validation plan. Do not fit a model yet.
- Read your context and instructions
- Compiled the causal ml starter
Proposed design
The target is the average treatment effect, ATE = E[Y(1)-Y(0)], for the stated study population. Confirm the sampling unit and the timing of each candidate control before finalizing the adjustment set.
Estimator
Use a cross-fitted IRM/AIPW design for this binary-treatment ATE. Fit the two conditional outcome functions and the treatment propensity in training folds; evaluate the augmented inverse-probability score on held-out observations. An unrestricted treatment-response model allows treatment effects to vary. A constant-slope PLR would add a different structural restriction and should be chosen deliberately.
Identification requirements
Document consistency, the relevant interference assumption, conditional exchangeability and overlap. Exclude controls affected by treatment. Statistical fitting does not establish these assumptions.
Validation plan
Inspect propensity support, treatment-arm sizes, nuisance out-of-fold losses and influential observations. If observations share a dependence unit, keep that unit intact in all learned preprocessing and folds, and use compatible uncertainty estimation. Define the analysis and learner choices before viewing effect significance.
Deliverable
This is a worked method-selection example, with no data fitted and no numerical treatment effect claimed. The next step is to confirm the design, then implement the selected score in an execution-capable module.
Connects securely to your tools. The creator never sees your data.
About this skill
Start with the research question
One complete skill, causal-method-router, helps your agent define the observational unit, outcome, treatment, controls and estimand before selecting an estimator.
Included
- A method map covering PLR, IRM/AIPW, S/T/X/DR learners, causal forests and IV designs.
- Identification and inference checklists, with equations and official source links.
- Guidance for continuous treatments, grouped data and feasible treatment contrasts.
Scope
This free tier provides design and method selection. Estimation scripts are included in Essentials and Complete. A causal interpretation still requires substantive assumptions and adequate support.
Original community synthesis grounded in Microsoft EconML and DoubleML. The full MIT source is public: https://github.com/Mat-Wong/Causal-ML-Skills
Method-selection guidance only. Identification assumptions require substantive evidence. No estimation scripts are included in Starter; theoretical coverage does not imply every method has a bundled executable implementation.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 8 days ago
- Free with an Agensi account
Requires
Permissions
Reads the included skill and references. Does not fetch private data, make API calls or execute estimation scripts in this tier.