Practical programming research of Linear DML model based on the simplest Python code: From the standpoint of novice researchers
Shunxin Yao
TL;DR
This work investigates the practicality of linear Double Machine Learning for causal inference using the simplest Python code in a Jupyter/Anaconda setup, focusing on novice users and API usability. It contrasts traditional causal methods with DML, outlines the theoretical framework including conditional expectation estimation and residualization, and empirically compares several base learners on synthetic data. Key findings reveal that while DML provides robust bias control, API usability and data-dimension issues hinder novice adoption, and model choice should align with goals such as speed (OLS/Ridge) or predictive accuracy (RF/XGBoost). The study highlights the need for more intuitive tooling and thorough tutorials to bridge the gap between DML theory and practical data analysis in Python. The results have practical implications for high-dimensional causal inference in economics and related fields, underscoring both the promise and the current limitations of accessible DML implementations.
Abstract
This paper presents linear DML models for causal inference using the simplest Python code on a Jupyter notebook based on an Anaconda platform and compares the performance of different DML models. The results show that current Library API technology is not yet sufficient to enable novice Python users to build qualified and high-quality DML models with the simplest coding approach. Novice users attempting to perform DML causal inference using Python still have to improve their mathematical and computer knowledge to adapt to more flexible DML programming. Additionally, the issue of mismatched outcome variable dimensions is also widespread when building linear DML models in Jupyter notebook.