Short-term forecasting using SARIMA models: an application to epidemiological time series data with COVID-19 in Kazakhstan as an example

Evaluating Time-Series Models for Dengue Haemorrhagic Fever Prediction: Arima vs. Sarima and Exponential Smoothing

Accurately forecasting the transmission of infectious diseases, such as Dengue Haemorrhagic Fever (DHF), is essential for robust public health planning. This study assesses the performance of three time-series forecasting models—Exponential Smoothing, Autoregressive Integrated Moving Average (ARIMA), and Seasonal ARIMA (SARIMA)—in predicting annual DHF cases in Malaysia from 2011 to 2021. Model accuracy was evaluated using the Mean Absolute Percentage Error (MAPE). Results indicate that ARIMA and SARIMA yielded the lowest error rates (73.34%), whereas Exponential Smoothing performed less effectively (76.79%). ARIMA was preferred due to its simplicity and comparable predictive power. Although these error rates exceed the standard 10–20% threshold for reliable forecasting, the study demonstrates that these models have significant potential when integrated with more granular data and external covariates. Furthermore, a three-segment piecewise nonlinear regression was employed to identify structural changes in DHF trends. This combined analytical approach provides deeper insights into disease progression and strengthens the evidence base for effective public health interventions.

Forecasting HIV Incidence Trajectories in Cebu City, Philippines: An ARIMA-Based Time-Series Analysis for Epidemic Surveillance and Public Health Planning

Background: The Philippines is experiencing one of the fastest-growing HIV epidemics in the Western Pacific region, with Cebu City among the most affected urban centers. However, localized forecasting studies remain limited, constraining evidence-based public health planning and resource allocation. This study aims to develop and evaluate an Autoregressive Integrated Moving Average (ARIMA) model for forecasting monthly HIV incidence in Cebu City. Methods: Monthly HIV surveillance data from January 2022 to March 2025 were analyzed using time-series forecasting. Candidate ARIMA models were assessed using the Akaike Information Criterion (AIC), parameter significance, residual diagnostics, and out-of-sample forecasting accuracy. Results: ARIMA (0,1,1) demonstrated the best predictive performance, achieving a Mean Absolute Percentage Error (MAPE) of 25.04% and outperforming Naïve, Simple Moving Average, and Exponential Smoothing models. Forecasts for April to December 2025 project monthly HIV cases to increase from 33 to 36. Although forecasts suggest a persistent rise in HIV incidence, widening confidence intervals indicate greater uncertainty over longer forecasting horizons. Conclusion: The projected increase in HIV incidence signals a continuing public health challenge in Cebu City. ARIMA-based forecasting provides a practical decision-support tool for anticipating disease trends and informing surveillance, targeted prevention, and healthcare resource allocation.

Analysis of Count Time Series Through Integer‐Valued Autoregressive (INAR) Process With Zero‐Inflation and Seasonality

In this article, we study a count time‐series data of weekly dengue cases in Kaohsiung City, Taiwan during the period 2009–2012, where the problems of a large number of zeros or zero‐inflation and seasonality arise together. To capture both features, we consider an application‐oriented zero‐inflated seasonal PINAR framework based on zero‐inflated Poisson innovations. We also incorporate an exogenous variable, namely the weekly maximum temperature, in the innovations to study the effect of temperature on the number of dengue cases as many studies have found that temperature is a significant factor in spreading dengue infections globally. The proposed model can capture both the non‐recovery cases from the previous time point and the new cases coming at the current time point. The distributional and forecasting properties of the proposed model are derived. The consistency and asymptotic normality of the CLS estimator are established under suitable regularity conditions, and simulation experiments are used to examine the finite‐sample estimation and forecasting performance of the proposed model. The proposed model is compared with some existing INAR models. Finally, we analyze the data of weekly dengue cases in Kaohsiung for practical illustration.

Second-order season-trend GRU for long time series forecasting

Machine Learning and ARIMA Model Averaging for Adaptive Public Health Forecasting: Comparative Evaluation and an Ontario COVID-19 Case Study

These findings support selecting forecasting models according to operating conditions rather than relying on a single universally preferred approach, and provide a practical framework for combining complementary statistical and machine-learning forecasts.

Forecasting and Financial Risk Analysis Using ARIMA Intervention Model: A Case Study on JMAS Tbk

The capital market is characterized by high volatility, making accurate forecasting and risk measurement essential for investment decision-making. This study aims to apply the Autoregressive Integrated Moving Average (ARIMA) intervention model and the Value at Risk (VaR) approach for stock prices forecasting and finansial risk analysis.The data consist of monthly stock prices of PT Asuransi Jiwa Syariah Jasa Mitra Abadi Tbk (JMAS) from January 2018 to December 2025. The ARIMA intervention model is applied to identify and quantify structural changes caused by external shocks, particularly the COVID-19 pandemic. The results indicate that the ARIMA (1,2,0) intervention model is the most appropriate model, with a step intervention function reflecting a sudden and permanent impact on stock price movements. The model satisfies diagnostic assumptions and demonstrates good forecasting accuracy, with a Mean Absolute Percentage Error (MAPE) of 13.01%. Forecasting results for January to March 2026 show that stock prices are expected to remain relatively low, indicating a slow post-pandemic recovery. Financial risk is measured using the Cornish-Fisher Value at Risk (VaR) approach at a 95% confidence level. The estimated VaR is -0.4245 indicating a maximum potential loss of 42.45% over a one-month period. This high level of risk reflects extreme market conditions during the COVID-19 period, which significantly increased volatility. This study contributes to financial analysis by integrating intervention-based time series modeling with risk measurement in a unified framework.