Stability and L2-penalty in Model Averaging

Authors: Hengkun Zhu, Guohua Zou

JMLR 2024 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the usefulness of the proposed method with a Monte Carlo simulation and application to a prediction task on the wage1 dataset. Keywords: Model averaging, Stability, Mean squared prediction error, L2-penalty
Researcher Affiliation Academia Hengkun Zhu EMAIL Guohua Zou EMAIL School of Mathematical Sciences Capital Normal University Beijing 100048, China
Pseudocode Yes Algorithm 1 Optimal weighting based on cross validation
Open Source Code No The paper mentions using "regsubsets in leaps package of R language" for one of its analyses, which refers to a third-party tool. However, it does not explicitly state that the authors' own code for the proposed methodology (L2-penalty model averaging) is being released or provide a link to it.
Open Datasets Yes In this section, we apply the proposed method to the real wage1 dataset in Wooldridge (2003) from the US Current Population Survey for the year 1976. There are 526 observations in this dataset.
Dataset Splits Yes We randomly divide the data into two parts: a training sample S of n observations for estimating the models and a test sample St of nt = 529 n observations for validating the results.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the simulations or real data analysis, such as GPU or CPU models, memory, or other computer specifications.
Software Dependencies No The paper mentions "R language" and the "leaps package" (Section 5) but does not specify any version numbers for these software components or any other libraries used.
Experiment Setup Yes For Algorithm 1, we set L = 100, B = 10, l = 50 and b = 5.