Co2 Concentration Forecasting In Smart Cities Using A Hybrid Arima-tft Model On Multivariate Time Series Iot Data

This article discusses the use of hybrid approaches that combine one or more models from one or more categories to address the shortcomings of each of the previously discussed single-model categories. Specifically, the article focuses on the use of hybrid approaches for CO emissions forecasting. Examples of hybrid approaches discussed include a framework integrating index decomposition analysis (IDA) along with ANNs and data envelopment analysis (DEA), a model combining a general regression neural network and scenario analysis, a pure deep-learning hybrid, an ensemble Adaptive Neuro-Fuzzy Inference System (ANFIS) learning method, a hybrid approach blending the results of nine different algorithms, a two-step hybrid method, and a single model integrating ANNs, RF, and particle swarm optimization (PSO).