Abstract
Predictions of the Air Quality Index (AQI) can provide information on air quality, aiding individuals in personal protection and environmental conservation, and facilitating policymakers in implementing measures to control air pollution. In the prediction of AQI time series, single-step forecasting tasks are less challenging. However, as the length of the forecast sequence increases, the prediction error tends to grow remarkably. Furthermore, tacking only univariate features as model inputs can lead to biases in forecasting outputs, as the models fail to capture the overall trends of the series based on a single feature. To address this issue, we developed a novel time series prediction model named Temporal feature Encoded Informer (TE-Informer) using the historical air pollution dataset from Yan’an City. This model incorporates attention mechanisms and periodic encoding, utilizing multiple pollutant series as feature inputs. It enriches the Informer architecture with more comprehensive temporal and global features to predict AQI series. The experimental results demonstrated that the Informer model, enhanced with periodic time encoding, has a significant advantage in extracting features from the input information. In the context of multi-step AQI time series forecasting, the model achieved an MSE (Mean Squared Error) of 24.8692 and an R2-score of 0.9793, outperforming other comparative models in all comprehensive metrics. Consequently, this model can effectively improve prediction accuracy and better capture important characteristic information within the AQI time series. The source code is available at https://github.com/msskx/TE-Informer.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.