Machine Learning-Guided Prediction of the Removal Efficiency of Pharmaceuticals and Person

Xunhao Wang's dataset contains 339 data points compiled from 101 peer-reviewed studies to predict the removal efficiency of pharmaceuticals and personal care products (PPCPs) by photocatalytic membranes. The data covers 74 PPCPs, 6 photocatalyst categories, and 9 membrane materials. The gradient boosting regression tree model trained on this data achieved an R² of 0.645.

Use Cases

• Predicting PPCP removal efficiency based on operational conditions like light time and initial concentration.
• Analyzing the influence of photocatalyst category on degradation performance.
• Identifying critical PPCP substructures, such as hydroxyl and heterocyclic groups, for removal.
• Guiding the design and optimization of photocatalytic membrane systems based on feature importance analysis.

Strengths

• Data is compiled from 101 peer-reviewed studies, suggesting a literature-review foundation.
• Dataset includes 339 data points covering 74 distinct PPCPs, 6 photocatalyst categories, and 9 membrane materials.
• The associated ML model achieved a reported R² of 0.645, MAE of 0.110, and RMSE of 0.151.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Dataset size is 53.5 KB, indicating a very limited scope.

Provenance

Source

figshare

Collection Method

Compiled from 101 peer-reviewed studies.

Freshness

Last updated 2026-05-16 01:43:26; freshness should be verified.