Adaptive Debiased Lasso: Online Inference for High-Dimensional Streaming Data

A methodological paper and supplementary materials for a novel online inference algorithm in high-dimensional generalized linear models. The work, authored by Ruijian Han and last updated on 2026-06-04, includes simulation experiments and a real-world application to spam email classification. The associated files are 1.1 MB in size and available under a CC-BY-4.0 license.

Use Cases

• Testing online debiasing procedures for regression coefficients based on the described adaptive stochastic gradient descent algorithm.
• Benchmarking computational efficiency in streaming data scenarios as demonstrated in the simulation experiments.
• Applying the Adaptive Debiased Lasso estimator to binary classification tasks, such as the spam email example mentioned.

Strengths

• Includes a real data application to spam email classification, providing a concrete use case.
• Method is described as operating in single-pass mode, which likely reduces time and space complexity.
• Supplementary materials are available for reproducing the work, as stated in the description.

Limitations

• The dataset itself (e.g., the spam email data) is not directly provided; the primary files are PDF and DOCX documents describing the method.
• Row count and column definitions for any example data are unknown, limiting suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Ruijian Han via figshare

Collection Method

Methodological development and simulation; real data application mentioned.

Freshness

Last updated 2026-06-04 15:50:30; freshness should be verified.