Building Energy Consumption Dataset with Derived Temporal and Weather Features: A Five-Year Hourly Record from a Commercial Building in South Africa

This dataset presents a comprehensive collection of hourly energy consumption data from a commercial office building (Building B) located in South Africa, spanning from January 1, 2016 to December 31, 2020. The dataset contains 43,848 hourly observations with 30 variables, including 12 original measurements and 18 engineered features. The original data includes hourly energy consumption (kWh) and meteorological variables sourced from NASA's MERRA-2 reanalysis dataset, including temperature (maximum, minimum, and average), relative humidity, precipitation, and solar radiation. To enhance the dataset's utility for energy modeling and analysis, we have created 18 derivative features including: - Temporal features: year, month, day, hour, day of week, season, weekend indicator - Lag variables: previous hour's energy consumption (ENERGY_lag1) and same hour from previous day (ENERGY_lag24) - Rolling statistics: 24-hour rolling mean - Weather-derived metrics: Heating Degree Hours (HDH), Cooling Degree Hours (CDH), temperature range, and energy intensity ratios This dataset is particularly valuable for developing predictive models, analyzing energy demand patterns, and understanding the relationship between weather conditions and building energy consumption in subtropical climates. All data are provided in CSV format with detailed metadata.