DETeCT: Data-Efficient low-carbon Technology Classification for multiclass Targets

ProjectDETeCT presents a data-driven framework to classify the adoption of low-carbon technologies (LCTs), such as electric vehicles, heat pumps, and solar PV, using high-frequency smart meter data. Combining domain-informed and time-series features within a Random Forest and stacking model, the approach achieves robust predictive performance on large-scale household data from Flanders. This data repository contains the complete model architecture, underlying data for training and testing the model, and complete results from the corresponding paper.Data sourcesAll data used in this project is published by Fluvius (2025) under an open data license that permits reproduction, redistribution and reuse (https://opendata.fluvius.be/p/licentieopendatafluvius/). Fluvius (2025). Verbruiksprofielen digitale elektriciteitsmeters: kwartierwaarden voor een volledig jaar. https://opendata.fluvius.be/explore/dataset/1_50-verbruiksprofielen-dm-elek-kwartierwaarden-voor-een-volledig-jaar/information/

ProjectDETeCT 提出了一个数据驱动框架，用于分类低碳技术（low-carbon technologies, LCTs）的应用采用情况，涵盖电动汽车、热泵与光伏太阳能（solar PV）等品类，该框架基于高频智能电表数据构建。
该框架结合领域先验特征与时序特征，采用随机森林（Random Forest）与堆叠模型（stacking），在来自比利时法兰德斯地区的大规模家庭用电数据集上取得了稳健的预测性能。
本数据仓库包含完整的模型架构、用于模型训练与测试的原始数据集，以及对应学术论文中的全部实验结果。

数据来源
本项目所用全部数据均由Fluvius（2025）依据开放数据许可发布，该许可允许数据的复制、再分发与二次使用（https://opendata.fluvius.be/p/licentieopendatafluvius/）。具体数据集引用信息如下：
Fluvius (2025). 数字电力电表用电负荷曲线：完整年度的15分钟间隔数据. https://opendata.fluvius.be/explore/dataset/1_50-verbruiksprofielen-dm-elek-kwartierwaarden-voor-een-volledig-jaar/information/