Model Performance on TravisTorrent Test Set Using 31 Features

31 clean pre-built features were used to evaluate model performance on the TravisTorrent test set. Lalit Narayan Mishra published this 5.5 KB Excel file on figshare in May 2026. The dataset's specific row count and column definitions are not provided in the metadata.

Use Cases

• Benchmarking predictive models based on the 31 pre-built features mentioned in the description
• Comparing algorithm performance on the TravisTorrent test set
• Analyzing the relationship between software engineering metrics and model outcomes

Strengths

• The dataset uses 31 clean pre-built features, providing a structured starting point for analysis.
• It is a small file (5.5 KB), making it quick to download and inspect.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Lalit Narayan Mishra via figshare

Collection Method

Likely derived from the TravisTorrent software engineering data corpus.

Freshness

Last updated 2026-05-27 17:39:28; freshness should be verified.