Short-term forecasting of PM10 and PM2.5 concentrations with Facebook's Prophet Model at the Belgrade-Zeleno brdo

• DOI: 10.15233/gfz.2023.40.7
• Publication Date: 2023-11-24T07:24:23Z
• Authors: Filip Arnaut; Vesna Cvetkov; Dragana Đurić; Mileva Samardžić-...
• Keywords: Imputation (statistics); Interpolation; Benchmark (surveying)

We demonstrate the use of Facebook's Prophet (usually just called Prophet) model for short-term air quality forecasting at Belgrade-Zeleno brdo monitoring station. To address missing data, we applied minimally-altering data distribution imputation techniques. Linear interpolation proved effective gaps (1–3 hours), hourly mean method mid-term (24–26 and Hermite polynomial long-term (132–148 hours). The most significant change was a 3.4% shift in skewness. Partitioning time series enabled detailed assessment model, with PM2.5 predictions being more precise than PM10. Using longest yielded absolute errors 6.5 μg/m3 PM10 2.7 PM2.5. Based on 173 forecasts, anticipate root-mean-square values under 6.26 9.99 50% cases. demonstrates several advantages yields satisfactory results. In future research, results obtained from will serve as benchmark other models. Additionally, is capable providing be utilized research.