StreamFuels: Continuously updated fuel sales datasets for forecasting, classification, and pattern analysis

Description of the data and file structureWe are providing 5 datasets in TSF (time series format).All files include the series name, start and end timestamps, product type, and the time series values.<b>NOTE:</b> These datasets have values ​​updated until May 2025. To obtain the updated datasets, consider using the StreamFuels package, available at https://pypi.org/project/streamfuels/Files and variablesFile: yearly_fuel_sales_by_state.tsf<b>Description:</b> comprises 216 time series capturing the yearly historical sales of eight fuel products – ethanol, regular gasoline (gasoline-r), aviation gasoline (gasoline-a), liquefied petroleum gas (LPG), aviation kerosene (kerosene-a), illuminating kerosene (kerosene-i), fuel oil, and diesel – across 27 Brazilian states. Among these time series, 135 have records dating back to 1947, while 27 series begin in 1953, another 27 in 1959, and the remaining 27 in 1980.File: monthly_oilgas_operations_by_state.tsf<b>Description:</b> comprises 76 time series capturing the monthly records of five types of industrial operations – production, reinjection, flaring, self-consumption, and availability – for three products: natural gas, petroleum, and natural gas liquid (NGL), across the 27 Brazilian states. Only natural gas includes all five operations, while petroleum and NGL are limited to the production operation. Of the total, 31 series started in 1997, while the remaining 45 date back to 2000.File: yearly_fuel_sales_by_city.tsf<b>Description:</b> comprises 29,282 time series capturing the yearly sales of eight fuels and asphalt (a petroleum derivative) across 5,325 Brazilian cities. Most of the series begins in 1990 and 1992. However, we have some recent series that began in 2018.File: fuel_type_classification.tsf<b>Description:</b> comprises 14,032 time series, each with a fixed length of 12 observations (i.e., one year of sales) and eight possible class labels. Among the five datasets, it is the only one that is labeled, with each series associated with a specific fuel type.File: monthly_fuel_sales_by_state.tsf<b>Description:</b> comprises 216 time series capturing the monthly historical sales of eight fuel products across Brazil’s 27 states. All the time series dating back to 1990.Code/softwareAll the codes to collect and preprocessing the data are available at https://github.com/lucas-castrow/streamfuels/Access informationOther publicly accessible locations of the data:https://github.com/lucas-castrow/datasets_streamfuelsData was derived from the following sources:https://www.gov.br/anp/en/https://www.gov.br/anp/pt-br/centrais-de-conteudo/dados-abertos/vendas-de-derivados-de-petroleo-e-biocombustiveis