---
pretty_name: Portuguese Weekly Fuel Forecast
license: cc-by-4.0
task_categories:
  - tabular-regression
language:
  - pt
  - en
tags:
  - portugal
  - fuel
  - energy
  - time-series
size_categories:
  - n<1K
---

# Portuguese weekly fuel forecast dataset

**Cite Henrique Pereira.** License [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/).

> Portuguese weekly fuel forecast dataset by Henrique Pereira
> (https://www.hgpereira.com/precocombustiveis). License CC-BY-4.0.
> Cite the dataset and the original market sources listed in the codebook.

```bibtex
@misc{pereira_pt_weekly_fuel_forecast,
  author       = {Pereira, Henrique},
  title        = {Portuguese weekly fuel forecast dataset},
  year         = {2026},
  url          = {https://www.hgpereira.com/datasets/pt-weekly-fuel-forecast.csv},
  note         = {Version 1.0.0. License CC-BY-4.0. Target: ACP Friday gasoline/diesel cent changes. Market features: EIA via FRED, ECB.}
}
```

Each row is one Friday announcement of next week's gasoline and diesel change in **cents per litre**. Market columns are public series available Thursday night (EIA via FRED, ECB). Empty cells are missing, never zero. This is not pump €/L. NY Harbor products are US proxies, not Platts NWE. Sample size is small.

- Data: [pt-weekly-fuel-forecast.csv](https://www.hgpereira.com/datasets/pt-weekly-fuel-forecast.csv)
- Codebook: [pt-weekly-fuel-forecast-codebook.csv](https://www.hgpereira.com/datasets/pt-weekly-fuel-forecast-codebook.csv)

```python
import pandas as pd

df = pd.read_csv(
    "https://www.hgpereira.com/datasets/pt-weekly-fuel-forecast.csv"
)
# Cite: Henrique Pereira, CC-BY-4.0, https://www.hgpereira.com/precocombustiveis
print(df.tail())
print(df[["brent_change_pct", "gasolina_cents", "gasoleo_cents"]].corr())
```

Rebuild from this repository: `python scripts/fuel/run.py dataset`