High Precision Causal Model Evaluation with Conditional Randomization
Authors: Chao Ma, Cheng Zhang
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Empirical studies demonstrate the improved of our estimator, highlighting its potential on achieving near-RCT performance. |
| Researcher Affiliation | Industry | Chao Ma Microsoft Research Cambridge, UK chao.ma@microsoft.com Cheng Zhang Microsoft Research Cambridge, UK cheng.zhang@microsoft.com |
| Pseudocode | No | The paper does not contain any pseudocode or explicitly labeled algorithm blocks. |
| Open Source Code | No | The paper states that 'All methods are implemented via Econ ML package' and provides a link to the Econ ML GitHub repository, but it does not state that the code for the methodology described in *this* paper is open source or publicly available. |
| Open Datasets | No | The paper uses 'synthetic datasets' (csuite datasets) generated based on structural causal models and also 'synthetic observational data' generated via a described process. It does not use or provide access information for a pre-existing publicly available dataset. |
| Dataset Splits | No | The paper does not explicitly provide details about training, validation, and test dataset splits with specific percentages, counts, or explicit statements about how data was partitioned for validation. |
| Hardware Specification | No | The paper does not provide any specific hardware details such as GPU or CPU models, or cloud computing specifications used for running the experiments. |
| Software Dependencies | No | The paper mentions 'All methods are implemented via Econ ML package' but does not provide specific version numbers for this package or any other software dependencies. |
| Experiment Setup | Yes | The paper details experimental settings such as varying the degree of imbalance (σ2 β = 1, 5, 10), the dimensionality of covariates (nx = 30, nw = 30), and repeating simulations '100 times'. |