Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Predicting the number of births in Skåne: a Time Series approach with external regressors

Syrimi, Chloe LU (2025) In Bachelor's Thesis in Mathematical Statistics MASK11 20251
Mathematical Statistics
Abstract
Effective workload planning in Skåne hospitals relies on accurate, data-driven birth forecasts. Weekly and hospital-level predictions are particularly important in the healthcare sector for operational decision-making, including staffing and maternity unit resource allocation. This study develops and evaluates a time series forecasting model capable of predicting weekly birth counts in the Skåne region. A regression model with seasonal ARIMA errors (regSARIMA) is implemented, incorporating lagged female population data across key reproductive age groups as external regressors. The model uses historical birth and population data from 2015 to 2024. Model robustness is evaluated through residual diagnostics and out-of-sample forecasting... (More)
Effective workload planning in Skåne hospitals relies on accurate, data-driven birth forecasts. Weekly and hospital-level predictions are particularly important in the healthcare sector for operational decision-making, including staffing and maternity unit resource allocation. This study develops and evaluates a time series forecasting model capable of predicting weekly birth counts in the Skåne region. A regression model with seasonal ARIMA errors (regSARIMA) is implemented, incorporating lagged female population data across key reproductive age groups as external regressors. The model uses historical birth and population data from 2015 to 2024. Model robustness is evaluated through residual diagnostics and out-of-sample forecasting accuracy (2025-2033), using projected population data. Evaluation metrics include Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and residual analysis using autocorrelation (ACF) and partial autocorrelation (PACF) functions, along with tests such as Ljung-Box (LB) and Jarque-Bera (JB). The proposed model requires estimation of just four parameters and achieves a MAE of 16.68 and an RMSE of 20.56, corresponding to a relative error of approximately 6% over the one-year test period. Residual diagnostics confirm that the model effectively captures seasonality and autocorrelation structures while remaining parsimonious. The analysis also highlights the dominant predictive role of the 25-29 age group and the delayed relationship between population structure and birth outcomes, consistent with pregnancy duration. Overall, the regSARIMA model demonstrates strong predictive performance and low computational cost, making it a suitable candidate for integration into regional healthcare planning systems. (Less)
Please use this url to cite or link to this publication:
author
Syrimi, Chloe LU
supervisor
organization
course
MASK11 20251
year
type
M2 - Bachelor Degree
subject
keywords
birth forecasting, time series analysis, regSARIMA, healthcare planning
publication/series
Bachelor's Thesis in Mathematical Statistics
report number
LUNFMS-4082-2025
ISSN
1654-6229
other publication id
2025:K14
language
English
id
9194104
date added to LUP
2025-06-11 15:40:02
date last changed
2025-06-11 15:40:02
@misc{9194104,
  abstract     = {{Effective workload planning in Skåne hospitals relies on accurate, data-driven birth forecasts. Weekly and hospital-level predictions are particularly important in the healthcare sector for operational decision-making, including staffing and maternity unit resource allocation. This study develops and evaluates a time series forecasting model capable of predicting weekly birth counts in the Skåne region. A regression model with seasonal ARIMA errors (regSARIMA) is implemented, incorporating lagged female population data across key reproductive age groups as external regressors. The model uses historical birth and population data from 2015 to 2024. Model robustness is evaluated through residual diagnostics and out-of-sample forecasting accuracy (2025-2033), using projected population data. Evaluation metrics include Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and residual analysis using autocorrelation (ACF) and partial autocorrelation (PACF) functions, along with tests such as Ljung-Box (LB) and Jarque-Bera (JB). The proposed model requires estimation of just four parameters and achieves a MAE of 16.68 and an RMSE of 20.56, corresponding to a relative error of approximately 6% over the one-year test period. Residual diagnostics confirm that the model effectively captures seasonality and autocorrelation structures while remaining parsimonious. The analysis also highlights the dominant predictive role of the 25-29 age group and the delayed relationship between population structure and birth outcomes, consistent with pregnancy duration. Overall, the regSARIMA model demonstrates strong predictive performance and low computational cost, making it a suitable candidate for integration into regional healthcare planning systems.}},
  author       = {{Syrimi, Chloe}},
  issn         = {{1654-6229}},
  language     = {{eng}},
  note         = {{Student Paper}},
  series       = {{Bachelor's Thesis in Mathematical Statistics}},
  title        = {{Predicting the number of births in Skåne: a Time Series approach with external regressors}},
  year         = {{2025}},
}