Abstract:Natural gas consumption forecasting is crucial for planning and operating of sustainable energy systems. The accuracy of consumption forecasting is significantly affected by the quality of the collected features. Previous feature clustering methods, such as K-means and Gaussian mixed model (GMM), ignore the interference of factors with weak correlation on the clustering effect and thus fail to extract key information from the collected features. This paper proposes a novel feature enhancement method, namely, Gaussian correlation mixed clustering (GCMC), to extract fluctuation patterns from the highest correlation factors and divides the original sequence into multiple clusters to enhance the feature quality while reducing the complex fluctuation. Among them, correlation coefficient analysis, GMM, Bayesian information criterion and an improved information evaluation method are combined to cluster the selected highest correlation feature based on fluctuation patterns and evaluate the enhancement effect of feature quality. Then, each of the divided clusters is regarded as an independent dataset of the long short-term memory (LSTM) model for parallel forecasting and the results are restored to the structure of original sequence. In our experiments, we design four real-life datasets with different complexities. The results reveal that the proposed method outperforms GMM in terms of information entropy and accuracy. The information entropy for evaluating feature quality is improved by 6.13–9.66%. In comparison with other classic forecasting models, the mean absolute range normalized error (MARNE) of GCMC-LSTM for Karditsa, Thessaloniki, Oinofyta and Salfa Anthoussa are 6.06%, 4.62%, 14.18% and 15.30%, respectively, which presents the best performance and robustness. Especially for datasets with high complexity, by introducing GCMC, the MARNE is improved by 32.72% in Oinofyta.

Demand Forecasting for Petrol Products in Gas Stations Using Clustering and Decision Tree.

A Clustering-Based Sales Forecast Method For Big Promotion Days In O2o On-Demand Retailing

Demand forecasting for production planning decision-making based on the new optimised fuzzy short time-series clustering

Gasoline Sales Prediction and Sample Data Design

Application of K Nearest Neighbor Clustering Based on the Combination Technology on Coal and Gas Outburst Forecasting

A New Hybrid Approach of Clustering Based Probabilistic Decision Tree to Forecast Wind Power on Large Scales

Input Data Selection for Daily Traffic Flow Forecasting Through Contextual Mining and Intra-Day Pattern Recognition

A new deep clustering method with application to customer selection for demand response program

Variable-weight Combined Forecasting Model with Causal Analysis and Clustering for Refined Oil Sales Forecasting

Application Of Clustering Technique To Electricity Customer Classification For Load Forecasting

Decision-making Oriented Clustering: Application to Pricing and Power Consumption Scheduling

A clustering-based feature enhancement method for short-term natural gas consumption forecasting

A novel short-term load forecasting framework based on time-series clustering and early classification algorithm

Data-driven Combining Forecasting Method for the Net Demand of Power Retailers in Distribution Network

Demand Forecast of Petroleum Product Consumption in the Chinese Transportation Industry

Cluster-based hierarchical demand forecasting for perishable goods

Sales forecasting of a food and beverage company using deep clustering frameworks

PurTreeClust: A Clustering Algorithm for Customer Segmentation from Massive Customer Transaction Data

Research on E-Commerce Retail Demand Forecasting Based on SARIMA Model and K-means Clustering Algorithm

Dynamic Model Selection Based on Demand Pattern Classification in Retail Sales Forecasting

A Method of Traffic Flow Forecast and Management