Python Code Examples for Causal Inference Models

📚 Quick Study Guide

• 🔬 Causal Inference: A process of determining the cause-and-effect relationships between variables.
• 🐍 Python Libraries: Key libraries include `causalml`, `DoWhy`, and `EconML`.
• 📊 Potential Outcomes: Framework introduced by Rubin, defining causality through potential outcomes under different treatments.
  • $Y_i(1)$: Outcome for individual $i$ if treated.
  • $Y_i(0)$: Outcome for individual $i$ if not treated.
• 🎯 Average Treatment Effect (ATE): The average difference in outcomes if everyone received the treatment versus if no one received the treatment. $ATE = E[Y(1) - Y(0)]$.
• 🚧 Assumptions: Key assumptions for causal inference include ignorability (no unobserved confounders), positivity (sufficient treatment variation), and consistency (well-defined treatment).
• ⚙️ Methods: Common methods include:
  • Propensity Score Matching (PSM): Matching treated and control units based on their propensity scores.
  • Inverse Probability of Treatment Weighting (IPTW): Weighting observations by the inverse of their probability of treatment.
  • Instrumental Variables (IV): Using an instrument to isolate the causal effect of the treatment.

Practice Quiz

1. Which Python library is commonly used for implementing causal inference models?
   1. A. TensorFlow
   2. B. PyTorch
   3. C. CausalML
   4. D. Scikit-learn
2. What does ATE stand for in the context of causal inference?
   1. A. Average Treatment Estimation
   2. B. Actual Treatment Effect
   3. C. Average Treatment Effect
   4. D. Adjusted Treatment Evaluation
3. Which of the following is a key assumption for causal inference?
   1. A. Perfect Multicollinearity
   2. B. Ignorability
   3. C. Heteroscedasticity
   4. D. Autocorrelation
4. What is Propensity Score Matching (PSM) used for?
   1. A. Predicting future outcomes
   2. B. Balancing covariates between treatment and control groups
   3. C. Feature selection
   4. D. Time series forecasting
5. What does the potential outcome $Y_i(1)$ represent?
   1. A. The outcome for individual $i$ if not treated
   2. B. The outcome for individual $i$ if treated
   3. C. The observed outcome for individual $i$
   4. D. The predicted outcome for individual $i$
6. Which method involves weighting observations by the inverse of their probability of treatment?
   1. A. Regression Analysis
   2. B. Instrumental Variables (IV)
   3. C. Inverse Probability of Treatment Weighting (IPTW)
   4. D. Propensity Score Matching (PSM)
7. What is the purpose of using Instrumental Variables (IV) in causal inference?
   1. A. To increase sample size
   2. B. To predict missing data
   3. C. To isolate the causal effect of the treatment
   4. D. To reduce computational complexity