IVMD-CNN-SVM: KDE Bandwidth Comparison for Hydrological Prediction Error Metrics

A 5.5 KB Excel file published on figshare by Hong Ma on March 17, 2026. It compares Kernel Density Estimation (KDE) bandwidths for a hybrid deep learning model (IVMD-CNN-SVM) applied to stationary runoff series. The data includes error fit metrics (EMA, ERMSE, ER2) and 90% prediction interval metrics (PICP, PINAW, F) for two hydrological stations.

Use Cases

• Comparing KDE bandwidth selection methods based on error fit and prediction interval metrics.
• Evaluating the performance of the IVMD-CNN-SVM hybrid model for stationary runoff series prediction.
• Benchmarking prediction interval reliability (PICP) and sharpness (PINAW) for hydrological forecasts.
• Analyzing the trade-off between prediction error distribution and interval width in machine learning models.

Strengths

• Dataset is small (5.5 KB), facilitating quick download and inspection.
• Explicitly includes multiple error metrics (EMA, ERMSE, ER2) and prediction interval metrics (PICP, PINAW, F).
• Released under a permissive CC-BY-4.0 license for open use and redistribution.

Limitations

• Row count and column-level documentation are absent; field semantics must be inferred after download.
• Data is limited to results from only two stations, which may limit generalizability.
• Description metadata is limited; actual data quality and context require manual inspection.

Provenance

Source

figshare, author Hong Ma.

Collection Method

Likely contains results from a computational experiment comparing KDE bandwidths for a specific machine learning model.

Freshness

Last updated 2026-03-17 17:37:34; freshness should be verified.