---
license: cc-by-4.0
task_categories:
- time-series-forecasting
- tabular-regression
tags:
- air-quality
- smart-city
- time-series
- regression
- gas-sensors
- pollution-monitoring
- uci
configs:
- config_name: default
  data_files:
  - split: train
    path: processed/air_quality.parquet
---

<p align="center">
----------------------------------------------------------------------------------------------------------------------------------------------------
</p>

<p align="center">
  <b>
    <span style="color:#2563eb;">Air quality monitoring</span>
    &nbsp;|&nbsp;
    <span style="color:#16a34a;">Time-series forecasting</span>
    &nbsp;|&nbsp;
    <span style="color:#9333ea;">Smart city research</span>
    &nbsp;|&nbsp;
    <span style="color:#ea580c;">Gas sensor analysis</span>
  </b>
</p>

<p align="center">
----------------------------------------------------------------------------------------------------------------------------------------------------
</p>

## 1. Project Introduction

This repository provides a Hugging Face friendly version of the **Air Quality Italian City Dataset** from the UCI Machine Learning Repository.

The dataset contains hourly air-quality measurements from a gas multisensor device deployed in a polluted urban area in an Italian city. It includes chemical sensor responses, reference analyzer measurements, temperature, relative humidity, and absolute humidity.

This dataset is useful for building projects related to:

- air quality prediction
- smart city monitoring
- pollution forecasting
- sensor drift analysis
- environmental digital twins
- LLM-assisted environmental data explanation

## 2. Dataset Summary

| Item | Description |
|---|---|
| Dataset type | multivariate time-series |
| Task | regression |
| Records | 9,358 hourly samples |
| Features | 15 variables |
| Time period | March 2004 to February 2005 |
| Missing values | tagged as `-200` in the original file |
| Source | UCI Machine Learning Repository |

## 3. Repository Structure

```text
Air-Quality-Italian-City/
├── README.md
├── data/
│   └── AirQualityUCI.xlsx
└── processed/
    └── air_quality.parquet
```

- `data/` contains the original Excel file.
- `processed/` contains the cleaned Parquet file for the Hugging Face Dataset Viewer.
- missing values marked as `-200` are converted to null values in the processed file.

## 4. Main Use Cases

| Use case | Description |
|---|---|
| **Air quality forecasting** | predict future pollution levels using historical sensor data |
| **Regression modeling** | estimate pollutant concentrations such as CO, NOx, NO2, or benzene |
| **Sensor analysis** | study chemical sensor behavior, drift, and cross-sensitivity |
| **Smart city research** | build pollution monitoring and environmental intelligence systems |
| **Digital twin project** | create a city air-quality digital twin for monitoring and explanation |

## 5. Column Dictionary

<table>
<tr>
<td valign="top">

| Column | Description |
|---|---|
| `Date` | date of measurement |
| `Time` | time of measurement |
| `CO_GT` | true CO concentration |
| `PT08_S1_CO` | CO-targeted sensor response |
| `NMHC_GT` | true non-methanic hydrocarbons concentration |
| `C6H6_GT` | true benzene concentration |
| `PT08_S2_NMHC` | NMHC-targeted sensor response |
| `NOx_GT` | true NOx concentration |

</td>
<td valign="top">

| Column | Description |
|---|---|
| `PT08_S3_NOx` | NOx-targeted sensor response |
| `NO2_GT` | true NO2 concentration |
| `PT08_S4_NO2` | NO2-targeted sensor response |
| `PT08_S5_O3` | O3-targeted sensor response |
| `T` | temperature |
| `RH` | relative humidity |
| `AH` | absolute humidity |
| `datetime` | combined date and time column |

</td>
</tr>
</table>

## 6. Loading the Dataset

```python
from datasets import load_dataset

dataset = load_dataset("SoyVitou/Air-Quality-Italian-City")

print(dataset)
print(dataset["train"][0])
```

## 7. Example Project: Air Quality Digital Twin

This dataset can be used to create a smart city air-quality digital twin.

```text
sensor data
    ↓
pollution forecasting model
    ↓
city air-quality digital twin
    ↓
LLM explanation assistant
    ↓
environmental recommendation
```

Example question:

```text
why did NO2 increase during this time period?
```

Example answer:

```text
NO2 increased because the reference analyzer value rose together with related sensor responses. Temperature and humidity may also affect sensor behavior, so these variables should be considered during analysis.
```

## 8. Suggested Machine Learning Tasks

| Task | Target |
|---|---|
| pollution regression | predict CO, NOx, NO2, or C6H6 |
| time-series forecasting | forecast future air-quality values |
| anomaly detection | detect unusual pollution or sensor readings |
| missing value handling | improve data quality and imputation |
| digital twin monitoring | track city air-quality state over time |

## 9. Original Source

This dataset is derived from the UCI Machine Learning Repository:

- Dataset: Air Quality
- Creator: Saverio Vito
- DOI: https://doi.org/10.24432/C59K5F
- Source: https://archive.ics.uci.edu/dataset/360/air+quality

## 10. Citation

If you use this dataset, please cite the original UCI dataset:

```text
Vito, S. (2008). Air Quality [Dataset].
UCI Machine Learning Repository.
https://doi.org/10.24432/C59K5F
```

## 11. Note

This repository keeps the original `.xlsx` file and provides a processed `.parquet` version to make the dataset easier to preview and use in Hugging Face workflows.