Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
A Direct Estimation of High Dimensional Stationary Vector Autoregressions
Authors: Fang Han, Huanran Lu, Han Liu
JMLR 2015 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Moreover, we provide some experiments on both synthetic and real-world equity data. We show that there are empirical advantages of our method over the lasso-type estimators in parameter estimation and forecasting. ... We conduct numerical experiments on both synthetic and real data to illustrate the effectiveness of our proposed method compared to the competing ones, as well as obtain more insights on the performance of the proposed method. |
| Researcher Affiliation | Academia | Fang Han EMAIL Department of Biostatistics Johns Hopkins University Baltimore, MD 21205, USA Huanran Lu EMAIL Han Liu EMAIL Department of Operations Research and Financial Engineering Princeton University Princeton, NJ 08544, USA |
| Pseudocode | No | The paper describes mathematical formulations for optimization problems (e.g., equations 10, 11, 13, 14) but does not present any structured pseudocode or algorithm blocks with step-by-step instructions in a code-like format. |
| Open Source Code | No | We use the R package glmnet (Friedman et al., 2010) for implementing the lasso method in Hsu et al. (2008), and the simplex algorithm for implementing ours. |
| Open Datasets | Yes | We conduct numerical experiments on both synthetic and real data to illustrate the effectiveness of our proposed method... For the real data, we further compare the three methods on the equity data collected from Yahoo! Finance. ...We collect the daily closing prices for 91 stocks that are consistently in the S&P 100 index between January 1, 2003 and January 1, 2008. |
| Dataset Splits | Yes | We set n1 and n2 to be two numbers (representing the length of training data and the number of replicates)... The tuning parameters for the three methods are selected using the cross-validation procedure outlined in Section 5.1 with n1 = T/2, n2 = T/2, and the lag p predetermined to be 1. ...for t = 1248, . . . , 1257, we select the data set EJt, , where we have Jt = {j : t 100 j t 1}, as the training set. |
| Hardware Specification | Yes | All experiments are conducted on a 2816-core Westmere/Ivybridge 2.67/2.5GHz Linux server with 17T memory, a cluster system with batch scheduling. |
| Software Dependencies | No | We use the R package glmnet (Friedman et al., 2010) for implementing the lasso method in Hsu et al. (2008), and the simplex algorithm for implementing ours. ...are generated using the flare package in R (Li et al., 2015). |
| Experiment Setup | Yes | We consider the settings where the time series length T varies from 50 to 100 and the dimension d varies from 50 to 200. We create the transition matrix A1 according to five different patterns: band, cluster, hub, random, and scale-free. ... We then rescale A1 such that we have A1 2 = 0.5. ...The tuning parameters for the three methods are selected using the cross-validation procedure outlined in Section 5.1 with n1 = T/2, n2 = T/2, and the lag p predetermined to be 1. |