Biodegradation QSAR Dataset with 10 Activity Classes

QSAR biodegradation data contains molecular descriptors for predicting environmental degradation potential. The dataset includes 2000 samples across 10 distinct activity classes, generated for machine learning benchmarking. It originates from the OpenML platform, likely derived from experimental or computational chemistry studies.

Use Cases

• Predict the biodegradation activity class from 100 molecular descriptor columns using multi-class classification.
• Analyze feature importance among the 100 descriptor columns to identify key molecular properties influencing biodegradability.
• Train and compare model performance on a stratified sample of 2000 rows to ensure class distribution is maintained.
• Use the dataset as a benchmark for evaluating algorithm robustness across 10 distinct biodegradation outcome categories.

Strengths

• Dataset contains 2000 samples, providing a substantial base for model training.
• Includes 100 feature columns, offering a detailed molecular descriptor space for analysis.
• Stratified sampling ensures balanced representation across the 10 target classes.

Limitations

• Specific molecular descriptor names and their chemical meanings are not provided, limiting interpretability.
• The dataset is a generated sample (seed_2), which may not represent the full original data distribution.
• Lack of metadata on experimental source, measurement methods, or data provenance reduces contextual understanding.

Provenance

Source

OpenML platform

Collection Method

Likely a generated or preprocessed sample from an original QSAR biodegradation study.