A Multi-Feature Dataset of Daily Electricity Demand with Meteorological, Holiday, and Special Event Attributes for Mirpur and Gulshan Zones of Dhaka, Bangladesh (2024–2025)
Description
This dataset contains 597 daily electricity demand records covering the continuous period from 1 January 2024 to 19 August 2025 for two representative Electricity Distribution Service Areas (EDSAs) in Dhaka, Bangladesh: Mirpur (23.8223°N, 90.3654°E), a predominantly residential area, and Gulshan (23.7925°N, 90.4078°E), a predominantly commercial area. Electricity demand data were obtained directly from the Dhaka Electric Supply Company PLC (DESCO) with official authorization. The two zones, about 5.4 km apart, were selected to enable direct comparison of residential and commercial consumption under near-identical environmental conditions. The data are provided as two CSV files (MirpurDataset.csv and GulshanDataset.csv), each with 597 rows (one per day) and 13 identical columns, prepared using the same collection, preprocessing, feature engineering, and quality-checking pipeline. The target variable, Total Demand, is the total daily electricity demand in MWh, obtained by summing DESCO's 24 hourly demand values for each area. Daily demand ranged from 2,340 to 6,608 MWh in Mirpur and from 2,022 to 6,332 MWh in Gulshan. Fewer than 1% of demand observations were missing and were imputed as the average of the preceding and following days. Each record includes six daily weather variables — maximum, minimum, and mean temperature (°C), rainfall (mm), maximum wind speed (m/s), and sunshine duration (hours) — retrieved from the Open-Meteo historical archive for a reference point between the two areas and joined by date. Calendar features comprise the day of the week; one of seven local climatic seasons (Winter, Spring, Summer, Monsoon, Late Monsoon, Autumn, Late Autumn); holiday flags covering weekly holidays (Friday–Saturday), public holidays, and government-declared special holidays, annotated from official Government of Bangladesh calendars; and the names of major national, cultural, and religious events (e.g., Eid-ul-Fitr, Durga Puja, Pohela Boishakh), with overlapping events separated by "|" and "None" otherwise. The dataset supports electricity load forecasting, energy demand modeling with exogenous variables, holiday and festival effect analysis, smart-grid analytics, explainable AI research, and comparative residential–commercial energy studies.
Files
Steps to reproduce
The dataset was prepared through a systematic data preparation process involving data collection, aggregation, integration, feature engineering, preprocessing, and quality checking. Electricity demand data were collected from two Electricity Distribution Service Areas (EDSAs) in Dhaka, Bangladesh, namely Mirpur and Gulshan, through Dhaka Electric Supply Company PLC (DESCO) with official permission for academic research and publication. Mirpur primarily represents a residential electricity consumption area, while Gulshan primarily represents a commercial area. The dataset covers the period from 1 January 2024 to 19 August 2025, with 597 daily observations for each service area. The original electricity demand records were available at an hourly resolution. For each date, the 24 hourly demand values were summed to obtain a single daily electricity demand value. Daily meteorological data for the same study period were obtained from the Open-Meteo Archive API. Six weather variables were incorporated into the dataset: maximum temperature, minimum temperature, mean temperature, rainfall, maximum wind speed, and sunshine duration. The weather records were matched with the electricity demand records using the observation date. This integration ensured that each daily electricity demand observation was associated with the corresponding weather conditions. The resulting dataset therefore combines electricity demand information with relevant meteorological factors at a daily temporal resolution. Additional temporal and contextual features were generated from the observation date. Day of the week and season were derived directly from the date, while public holidays, major national and religious events, and special holidays were added using official Government of Bangladesh holiday calendars and relevant public event schedules. The integrated dataset was then checked for missing values, duplicate records, inconsistent dates, and data consistency across the two service areas. The final dataset was organized into a structured tabular format containing daily electricity demand, meteorological, temporal, holiday, and special event attributes for Mirpur and Gulshan. These steps make the dataset suitable for electricity demand analysis, forecasting, machine learning, and investigation of the effects of weather and social events on electricity consumption.
Institutions
- Bangladesh University of Business and TechnologyDhaka Division, Dhaka