PREDICTIVE MODELING IN SMART SITIES BASED ON ARIMA MODELS

DOI: 10.31673/2786-8362.2025.025360

Abstract

Accurate time series forecasting is a cornerstone of smart city
management, enabling informed decision-making across energy systems, transportation networks,
environmental monitoring, and public safety. This study investigates the application of ARIMA models for
urban time series prediction, focusing on both standard ARIMA and residual-corrected ARIMA variants.
Residual correction allows the capture of latent structural patterns and systematic deviations not fully
addressed by the base ARIMA model, thereby enhancing forecast accuracy. Model performance is
rigorously evaluated using a comprehensive set of metrics, including Mean Absolute Error (MAE), Root
Mean Squared Error (RMSE), symmetric Mean Absolute Percentage Error (sMAPE), Mean Absolute
Percentage Error (MAPE), and Mean Absolute Scaled Error (MASE), complemented by a multi-criteria
ranking procedure to identify the optimal model configuration. The results demonstrate that residualcorrected ARIMA consistently outperforms standard ARIMA for short- and medium-term forecasting
tasks, particularly for variables such as air temperature, which exhibit complex temporal dynamics. The
findings underscore the practical relevance of ARIMA-based forecasting as a reliable tool for data-driven
decision support in smart city infrastructures.
Keywords: time series, forecasting, ARIMA model, ARIMA with residual correction, model
selection criteria

References
1. Zhang Z., Ren S., Qian X., Duffield N. Towards Invariant Time Series Forecasting in Smart
Cities // arXiv preprint arXiv:2405.05430. 2024. URL: https://arxiv.org/abs/2405.05430.
2. Liang S. Predicting short-term urban bike sharing demand in a smart city // Computers,
Environment and Urban Systems. 2026. Vol. 88. P. 101703. URL:
https://doi.org/10.1016/j.compenvurbsys.2021.101703.
3. Box G.E.P., Jenkins G.M. Time Series Analysis: Forecasting and Control. Prentice Hall, 1970.
4. Hyndman R.J., Athanasopoulos G. Forecasting: Principles and Practice. OTexts, 2021. URL:
https://otexts.com/fpp3/.
5. Yavuz V.S. Forecasting monthly rainfall and temperature patterns in Van Province, Türkiye
using ARIMA and SARIMA models // Journal of Water and Climate Change. 2025.
6. GeeksforGeeks. ARIMA vs. SARIMA model. URL:
https://www.geeksforgeeks.org/machine-learning/arima-vs-sarima-model/.
7. Machine Learning Mastery. Model residual errors: Correct time series forecasts in Python.
URL: https://machinelearningmastery.com/model-residual-errors-correct-time-series-forecastspython/.
8. ARIMA-model of Ukrainian Macroeconomic Indicators Forecasting // In: Emergence of
public development: financial and legal aspects: monograph. Coventry: Agenda Publishing House,
2019. P. 213-221.
9. Application of residual modification approach in seasonal ARIMA for electricity demand
forecasting: A case study of China / Y. Wang et al. Energy Policy. 2012. Vol. 48. P. 284–294. URL:
https://doi.org/10.1016/j.enpol.2012.05.026.
10. Real-Time Temperature Prediction Models for Smart City Applications. URL:
https://www.researchgate.net/publication/388194104_RealTime_Temperature_Prediction_Models_for_Smart_City_Applications.
11. Kyiv, Ukraine Weather Statistics Dataset. URL:
https://www.kaggle.com/datasets/thedevastator/kyiv-ukraine-weather-statistics
12. APXML. ARIMA Model Diagnostics. URL: https://apxml.com/courses/time-seriesanalysis-forecasting/chapter-4-arima-models-forecasting/arima-modeldiagnostics?utm_source=chatgpt.com.
13. Bhutani K. Evaluating Time Series Models for Real-World Forecasting: A Practical
Comparison. URL: https://medium.com/%40karanbhutani477/evaluating-time-series-models-forreal-world-forecasting-a-practical-comparison-5c9622618715.
14. Khudov H., Makoveychuk O., Butko I., Khizhnyak I. A model for prediction of geospatial
data in systems for processing geospatial information // Systems of Arms and Military Equipment. –
2021. № 2(66). С. 123-128. URL: https://doi.org/10.30748/soivt.2021.66.16.
15. Google Colab. URL: https://colab.google/.
16. Pandas. URL: https://pandas.pydata.org/.
17. NumPy. URL: https://numpy.org/.
18. Statsmodels. URL: https://www.statsmodels.org/stable/index.html.
19. Seaborn. URL: https://seaborn.pydata.org/.