Data from: Improving natural gas demand forecasts: An approach combining bootstrap aggregation, time series methods and regularization techniques

Original data: monthly gross inland natural gas consumption (in Terajoules, TJ) in each of the 14 countries included in the analysis. The data were collected from the Statistical Office of the European Union (EUROSTAT, 2019) database. The time period of the analysis spans from January 2008 to June 2018 (the last date available for all European countries). Data ranging from January 2008 to June 2017 were considered as training set for models using the traditional aggregation routines (mean and median). For models employing the regularization routines, a validation set is included between the months of July 2016 and June 2017, in which appropriate weights are assigned to each of the bagged forecasts. The test set, in turn, comprised the last 12 observations (July 2017 – June 2018). The original data can also be found in the supplementary material of the submitted manuscript (Appendix C). A total of 14 countries were selected for the empirical analysis. An endeavor was made to include most of the major consuming countries in Europe, namely France, Germany, Italy, Netherlands, Turkey and United Kingdom. To demonstrate the flexibility of the proposed approaches, we also selected several medium (Belgium, Poland and Romania) and minor consuming countries (Denmark, Finland, Ireland, Luxembourg and Slovakia). We finally emphasize that not every country could be included in the analysis, due to the lack of data (missing information and/or dataset starting after January 2008). References: EUROSTAT (European Statistics) [dataset]. January 7, 2019 Supply of gas - monthly data [nrg_103m] / UNIT: Terajoules (gross calorific value = GCV) / PRODUCT: Natural gas / INDIC_NRG: Gross inland consumption. 2019 Jan 7 [cited 2019 Jan 7]. Available from: https://ec.europa.eu/eurostat/web/energy/data/database原始数据为本次分析涵盖的14个国家的月度内陆天然气总消费量（单位：太拉焦耳（Terajoules，TJ））。 该数据采集自欧盟统计局（EUROSTAT, 2019）数据库。本次分析的时间跨度为2008年1月至2018年6月（为所有欧洲国家可获取的最新有效截止日期）。其中，采用传统聚合方法（均值、中位数）的模型，其训练集为2008年1月至2017年6月的数据；针对采用正则化方法的模型，则额外设置了2016年7月至2017年6月的验证集，该集合将为每个装袋（bagged）预测分配合适权重。测试集则为最后12条观测数据，即2017年7月至2018年6月。原始数据亦可在已投稿论文的补充材料（附录C）中获取。 本次实证分析共选取14个国家，旨在覆盖欧洲多数主要天然气消费国，包括法国、德国、意大利、荷兰、土耳其及英国。为展示所提方法的灵活性，研究同时选取了若干中等消费规模国家（比利时、波兰、罗马尼亚）与小型消费国（丹麦、芬兰、爱尔兰、卢森堡及斯洛伐克）。需特别说明的是，由于部分国家存在数据缺失问题（信息不全或数据集起始时间晚于2008年1月），无法将其全部纳入本次分析。 参考文献： 欧盟统计局（European Statistics）[数据集]. 2019年1月7日，天然气供应——月度数据[nrg_103m] / 单位：太拉焦耳（总热值=GCV） / 产品：天然气 / 能源指标（INDIC_NRG）：内陆总消费量. 2019年1月7日 [引用日期2019年1月7日]. 可从以下网址获取：https://ec.europa.eu/eurostat/web/energy/data/database