Analyze the energy consumption characteristics and affecting factors of Taiwan's convenience stores-using the big data mining approach

•Machine learning is tolerant of complex convenience stores dynamic energy consuming.•The WEKA can explore the energy consumption characteristic and implicit knowledge.•The lighting and refrigeration are the key factors affecting energy consumption.•The minimum relative humidity is the key climate factor affecting energy consumption.•Configuring of lighting, area and refrigerator can optimize energy efficiency. This study applies big data mining, machine learning analysis technique and uses the Waikato Environment for Knowledge Analysis (WEKA) as a tool to discuss the convenience stores energy consumption performance in Taiwan which consists of (a). Influential factors of architectural space environment and geographical conditions; (b). Influential factors of management type; (c). Influential factors of business equipment; (d). Influential factors of local climatic conditions; (e). Influential factors of service area socioeconomic conditions. The survey data of 1,052 chain convenience stores belong to 7-Eleven, Family Mart and Hi-Life groups by Taiwan Architecture and Building Center (TABC) in 2014. The implicit knowledge will be explored in order to improve the traditional analysis technique which is unlikely to build a model for complex, inexact and uncertain dynamic energy consumption system for convenience stores. The analysis process comprises of (a). Problem definition and objective setting; (b). Data source selection; (c). Data collection; (d). Data preprocessing/preparation; (e). Data attributes selection; (f). Data mining and model construction; (g). Results analysis and evaluation; (h). Knowledge discovery and dissemination. The key factors influencing the convenience stores energy consumption and the influence intensity order can be explored by data attributes selection. The numerical prediction model for energy consumption is built by applying regression analysis and classification techniques. The optimization thresholds of various influential factors are obtained. The different cluster data are compared by using clustering analysis to verify the correlation between the factors influencing the convenience stores energy consumption characteristic. The implicit knowledge of energy consumption characteristic obtained by the aforesaid analysis can be used to (a). Provide the owners with accurate predicted energy consumption performance to optimize architectural space, business equipment and operations management mode; (b). The design planners can o