Universal Journal of Public Health Vol. 13(3), pp. 551 - 562
DOI: 10.13189/ujph.2025.130303
Reprint (PDF) (1181Kb)

Predictive and Time Series Analysis of Food Security Indices: Multiple Regression and ARIMA Models, Case Study of Morocco

Mehdi Rahmaoui ^1,*, Achraf Chakir Baraka ², Ahssaine Bourakadi ³, Nada Yamoul ⁴, Hamid Khalifi ², Abdellatif Bour ^1
1 Laboratory of Biology and Health, Team of Nutritional Sciences, Food and Health, Faculty of Sciences, Ibn Tofail University, Kenitra, Morocco
² Faculty of Sciences, Mohammed V University in Rabat, Morocco
³ Department of Mathematics, Faculty of Sciences, Ibn Tofail University, Kenitra, Morocco
⁴ Department of Physics, Faculty of Sciences, Ibn Tofail University, Kenitra, Morocco

ABSTRACT

The analysis of food security indices in Morocco enabled the study to be specified in terms of forecasting the prevalence of undernourishment as a % of the population in Morocco during the period between 2001 and 2002, while also analyzing the post and past COVID-19 periods. In this context, the modeling of the prevalence of undernourishment as a % of the population using explanatory variables such as the inflation rate and the consumer price index (CPI) produced a very powerful model, characterized by a very high correlation coefficient of 95.2%, in addition to the very high significance of the explanatory variables as demonstrated by statistical tests. Moreover, the predictions were generated using an ARIMA model, specifically an AR(2) model. The ARIMA (AutoRegressive Integrated Moving Average) model is able to forecast time series by combining three components: AR (regression on past values), I (differentiation to stabilize the series), and MA (incorporating previous forecast errors). Consequently, these predictions rely on correlated data. The research emphasizes the significance of methodological decisions when evaluating the prevalence of undernourishment, as a percentage of the population, in Morocco. By comparing, the two approaches: multiple regression models, and ARIMA time series models, the results obtained indicate forecasts of 8.69% and 5.91% respectively for the year 2024. The notable difference between the two forecasts highlights the importance of understanding the context and limitations of each model. The 8.69% forecast of the multiple regression models could reflect expectations based on external factors of concern for the year 2024, while the lower 5.91% forecast of the time series may mean that the observed trend is more optimistic. All the data and results come respectively from the World Bank's official website and the R software.

KEYWORDS
Prevalence of Undernourishment in % of the Population, Food Security, Modeling, Multiple Regression Models, Covid-19 Pandemic, Time Series Forecasting, ARIMA Time Series Models

Cite This Paper in IEEE or APA Citation Styles
(a). IEEE Format:
[1] Mehdi Rahmaoui , Achraf Chakir Baraka , Ahssaine Bourakadi , Nada Yamoul , Hamid Khalifi , Abdellatif Bour , "Predictive and Time Series Analysis of Food Security Indices: Multiple Regression and ARIMA Models, Case Study of Morocco," Universal Journal of Public Health, Vol. 13, No. 3, pp. 551 - 562, 2025. DOI: 10.13189/ujph.2025.130303.

(b). APA Format:
Mehdi Rahmaoui , Achraf Chakir Baraka , Ahssaine Bourakadi , Nada Yamoul , Hamid Khalifi , Abdellatif Bour (2025). Predictive and Time Series Analysis of Food Security Indices: Multiple Regression and ARIMA Models, Case Study of Morocco. Universal Journal of Public Health, 13(3), 551 - 562. DOI: 10.13189/ujph.2025.130303.