Causal Effect of Functional Treatment

Authors: Ruoxu Tan, Wei Huang, Zheng Zhang, Guosheng Yin

JMLR 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We study their theoretical properties, which are corroborated through extensive numerical experiments. A real data application on electroencephalography data and disease severity demonstrates the practical value of our methods. Simulation experiments and the real data analysis are presented in Sections 6 and 7, respectively.
Researcher Affiliation Academia Ruoxu Tan EMAIL School of Mathematical Sciences and School of Economics and Management Tongji University Shanghai, China; Wei Huang EMAIL School of Mathematics and Statistics University of Melbourne Melbourne, VIC, Australia; Zheng Zhang EMAIL Center for Applied Statistics, Institute of Statistics & Big Data Renmin University of China Beijing, China; Guosheng Yin EMAIL Department of Statistics and Actuarial Science University of Hong Kong Hong Kong SAR, China
Pseudocode No The paper describes methods verbally and mathematically (e.g., the backfitting algorithm for OR estimators), but does not contain a clearly labeled pseudocode or algorithm block.
Open Source Code Yes The code to reproduce the simulation results for our methods and the FLR is available at https://github.com/ruoxut/FunctionalTreatment.
Open Datasets Yes We illustrate the estimation of the ADRF using the five methods, FSW, OR, DR, PCW and FLR, on the electroencephalography (EEG) dataset from Ciarleglio et al. (2022).
Dataset Splits Yes In the first stage, we select the truncation parameter q using the OR estimator. Specifically, we randomly split the dataset into L parts, S1, . . . , SL. Let S ℓdenote the remaining sample with Sℓexcluded.
Hardware Specification Yes The experiments were carried out on a PC with an i7-12700 CPU and 16 GB RAM.
Software Dependencies No The paper mentions that code is available on GitHub but does not explicitly state specific software dependencies with version numbers, such as Python 3.8 or PyTorch 1.9.
Experiment Setup Yes We select the tuning parameters h, k and q following the procedure in Section 5, and choose the number of PCs used in estimating the PCW so as to explain 95% of the variance of Z following Zhang et al. (2021). For a fair comparison, we use the truncation parameter bq CV in Section 5 for estimating b in all methods that utilize the functional linear model. We consider the sample sizes n = 200 and 500, and generate data for models (i) to (v) as follows. where ϵi1 N(0, 1) and ϵi2 N(0, 25) are generated independently.