MARV Model Results: Protein Binding Predictions with Bootstrap Analysis

Model results for predicting the binding strength of the MARV protein to bat NPC1. The dataset includes evaluations for 500 parameter combinations, metrics for 100 bootstrap iterations of a chosen parameter set, and variable importance and SHAP values for each variable. The 59.2 MB collection of CSV files was authored by Adrian Castellanos and last updated in May 2026.

Use Cases

• Validate and compare model performance based on evaluation metrics for 100 bootstrap iterations.
• Analyze feature importance for protein binding predictions based on variable importance results.
• Interpret model predictions for individual variables using the provided SHAP values.
• Perform hyperparameter tuning by evaluating results across 500 different parameter combinations.

Strengths

• Includes results from 500 distinct parameter combinations for model evaluation.
• Provides 100 bootstrap iterations for robust assessment of a chosen parameter set.
• Contains SHAP values for model interpretability across 100 bootstrap iterations.

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

figshare

Collection Method

Computational model results, likely from a machine learning or statistical model.

Freshness

Last updated 2026-05-20 13:41:37; freshness should be verified.