Raman Spectroscopy Data for Aviation Fuel Property Prediction

A curated set of 160 liquids, including neat hydrocarbons, mixtures, and jet fuels, was measured with a compact Raman system. The dataset supports a two-tier chemometric framework for predicting derived cetane number and density, developed by Dhananjay Ambre and last updated in March 2026.

Use Cases

• Predict derived cetane number (DCN) based on Raman spectral features
• Estimate liquid density based on predicted UNIFAC composition vectors
• Train multitarget artificial neural networks for functional-group composition regression
• Perform Monte Carlo sensitivity analysis on spectral noise robustness
• Conduct permutation-importance analysis for structure-property trend validation

Strengths

• Contains Raman spectroscopy measurements for 160 curated liquids
• Tier-1 model reproduces functional-group compositions with test-set R² ≥ 0.95 for all UNIFAC groups
• DCN model attains test R² ≈ 0.94, with ∼74% of samples within ±10% of experimental measurements
• Density model achieves test-set R² = 0.90, with 91% of samples within ±2% of experimental measurements

Limitations

• Row count and column-level documentation are unknown, which may limit suitability assessment
• Dataset size is 447.8 KB, indicating a relatively small scope
• Data may reflect bias inherent to the specific sample collection and measurement methodology

Provenance

Source

figshare

Collection Method

Raman spectroscopy measurements of a curated set of liquids, with literature-guided peak assignments.

Freshness

Last updated 2026-03-19 05:45:51