Multivariate Air Quality Data for One-Step-Ahead AQI Forecasting

A dataset for short-term forecasting of the Air Quality Index (AQI) using a Transformer encoder model. Created by Kun-Chou Lee and last updated in May 2026, it contains multivariate time-series data where input features include AQI and other numerical variables. The dataset is preprocessed with forward-fill for missing values and Min-Max normalization, and supervised learning samples are constructed using a sliding-window scheme.

Use Cases

• Training one-step-ahead forecasting models based on multivariate air quality data.
• Benchmarking Transformer encoder architectures against other time-series models based on the described methodology.
• Developing real-time environmental decision-support systems based on short-term AQI predictions.
• Studying the impact of historical observation windows on forecast accuracy based on the five-step window mentioned.

Strengths

• Preprocessing steps are explicitly documented, including forward-fill for missing values and Min-Max normalization to prevent data leakage.
• Model performance on the test set is quantified with specific metrics: MSE = 19.84, RMSE = 4.45, and MAE = 3.57.
• The dataset is shared under a permissive CC-BY-4.0 license.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The geographic and temporal coverage of the underlying monitoring data is not specified.

Provenance

Source

figshare

Collection Method

Data preprocessing includes converting observation time to a temporal index, handling missing values, and applying normalization.

Freshness

Last updated 2026-05-13 05:51:23; freshness should be verified.