Fish Toxicity Data for QSAR Modeling

A Quantitative Structure-Activity Relationship dataset for predicting chemical toxicity to fish. The dataset is hosted by the UCI Machine Learning Repository and relates molecular descriptors to a toxicity endpoint. It is used for modeling in environmental chemistry and toxicology.

Use Cases

• Predict fish toxicity endpoint from molecular descriptors using regression models.
• Identify key molecular features influencing toxicity through feature importance analysis.
• Compare model performance between linear regression and non-linear methods like random forests on this benchmark dataset.
• Assess chemical risk by classifying compounds into toxicity categories based on the continuous toxicity value.

Strengths

• Curated and hosted by the UCI Machine Learning Repository, a trusted source for benchmark datasets.
• Provides a clear task for QSAR modeling with defined molecular features and a toxicity target.

Limitations

• The specific number of compounds (rows) and molecular descriptors (columns) is unknown.
• The temporal and geographic context of the original experimental data is unspecified.

Provenance

Source

UCI Machine Learning Repository

Collection Method

Compiled from experimental toxicology studies, likely involving the measurement of chemical toxicity to fish.