Forecasting the population of children within the daycare age in Lappeenranta
Tiony, Abigael Jemutai (2018)
Diplomityö
Tiony, Abigael Jemutai
2018
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe2018052524756
https://urn.fi/URN:NBN:fi-fe2018052524756
Tiivistelmä
This research uses the new efficient Markov chain Monte Carlo (MCMC) simulation method for statistical analysis that was presented by Marko Laine, the author of a MATLAB toolbox that we used to build simulation models [1].
In this study, we used MCMC to forecast the number of children within the daycare age bracket. When using MCMC methods, the model is simulated repeatedly to explore the probability distribution describing the uncertainties in model parameters and predictions and this is precisely what we need. We tested three different approaches such that each approach consisted two or three models that used MCMC to approximate parameters and compute a predictive distribution that shows the spread of uncertainty across the forecast period. The parameters of interest were fertility rate, survival rate and net immigration rates.
The results from the MCMC simulation revealed that in all the three approaches that we used in our experiment, model m3a from the first approach gave results with a lower uncertainty level. This is because the model error variance of this model was lower than the rest, and this makes its results more reliable. The results of this research gives a more realistic indication of the level of uncertainty that one should expect in the population size of childcare aged children in the next 20 years.
In this study, we used MCMC to forecast the number of children within the daycare age bracket. When using MCMC methods, the model is simulated repeatedly to explore the probability distribution describing the uncertainties in model parameters and predictions and this is precisely what we need. We tested three different approaches such that each approach consisted two or three models that used MCMC to approximate parameters and compute a predictive distribution that shows the spread of uncertainty across the forecast period. The parameters of interest were fertility rate, survival rate and net immigration rates.
The results from the MCMC simulation revealed that in all the three approaches that we used in our experiment, model m3a from the first approach gave results with a lower uncertainty level. This is because the model error variance of this model was lower than the rest, and this makes its results more reliable. The results of this research gives a more realistic indication of the level of uncertainty that one should expect in the population size of childcare aged children in the next 20 years.