Browsing by Subject "prediction"

Oil spills in aquatic environments are devastating disasters with both biological and economic impacts. Fish populations are among the many subjects of these impacts. In literature, there are numerous assessments of oil spill impacts on fish populations. From all applied research methods, the focus of this thesis is on Bayesian methods. In prior research, several Bayesian models have been developed for assessing oil spill impacts on fish populations. These models, however, have focused on the assessment of impacts from past spills. They have not been used for predicting impacts of possible future oil spills. Furthermore, the models have not utilized data from laboratory studies. Some examples can be found of models assessing economic impacts of oil spills on fish populations however, none of them assess the economic impacts that follow from decreases in biomass. The aim of this thesis is to develop a Bayesian bioeconomic prediction model, which would be able to predict oil spill impacts on Baltic Sea main basin herring population, and the consequential economic impacts on fishermen. The idea is to predict the impacts of several hypothetical oil spill scenarios. As a result of this thesis, a bioeconomic prediction model was developed, which can predict both biological and economic impacts of oil spills on Baltic Sea main basin herring through additional oil induced mortality of herring eggs. The model can be applied to other fish populations in other regions as well. The model utilizes laboratory studies for assessing population level impacts. The model can be used for both assessing risks of the impacts of possible future oil spills, and for decision analysis after a spill has already occurred. Furthermore, the model can be used for assessing unknown aspects of past oil spills. The economic predictions can be used, for example, to estimate the compensations that could possibly be paid to fishermen. In the future, the prediction model should be developed further, especially regarding its stock-recruitment relationship assumptions. In addition, the model’s assumptions regarding the calculation of oil induced additional mortality and the economic impacts, should be expanded.

Itsemurha on edelleen tärkeä kuolinsyy Suomessa. Itsemurhariskissä olevan potilaan tunnistaminen ja itsemurhien ennaltaehkäisy on haastavaa. Viime aikoina Implisiittistä assosiaatiotestiä (IAT) on yhä enemmän sovellettu psykologisissa ja psykiatrisissa tutkimuksissa, joissa on saatu näyttöä psykiatristen potilaiden automatisoituneiden mielleyhtymien ja itsetuhokäyttäytymisen yhteydestä. Halusimme tutkimuksellamme selvittää, onko aiemmin itsemurhaa yrittäneillä suomalaisilla masennuspotilailla vahvempi implisiittinen assosiaatio itsensä ja kuoleman/itsemurhan välillä, kuin potilailla, joilla ei ole itsemurhayritystä taustalla. Rekrytoimme tutkimukseen yhteensä 31 masennuspotilasta HYKS Mielialalinjan aluepoliklinikoilta ja vanhuspsykiatrian avohoidosta. Tutkittavista 8 oli aikaisemmin yrittänyt itsemurhaa. IAT:n reaktioaikojen perusteella jokaiselle tutkittavalle laskettiin D-arvo osoitukseksi implisiittisestä assosiaatiosta itsensä ja itsemurhan/kuoleman välillä. Negatiivinen D-arvo osoittaa vahvempaa assosiaatiota itsensä ja elämän välillä, kun taas positiivinen D-arvo osoittaa vahvempaa assosiaatiota itsemurhaan/kuolemaan. Ryhmien välisiä suorituksia verrattiin T-testin avulla. Tuloksissa molempien ryhmien keskimääräinen D-arvo oli negatiivinen. Itsemurhaa yrittäneiden D-arvo oli kuitenkin keskimäärin vähemmän negatiivinen kuin verrokkiryhmän. Ero ei ollut tilastollisesti merkittävä N-määrällä 31. Tutkimuksemme aineisto ei tukenut käsitystä IAT:n hyödyllisyydestä itsemurhariskin arvioinnissa. Lopulliseen päätelmään menetelmän hyödyllisyydestä kliinisessä työssä tarvitaan vielä lisää replikaatiotutkimuksia.

There develops heartwood in the stems of the Scots pines (Pinus sylvestris L.) that differs by its natural characteristics from the other sections of the wood material in the pine stem. Pine heartwood is natural-ly decay resistant and it can be used in conditions where the normal wood products can’t be used. The aim of this study was to develop a method, which can be used for predicting the diameter and volume of heartwood. There is a need for this kind of method, because it still is not possible to estimate the amount of heartwood in a standing tree without damaging the tree itself. The variables measured from single trees describing the diameter of the heartwood on eight relative heights were analysed by using linear regression. When the best explanatory variables were selected, a mixed linear model was created for each of the relative heights. The mixed linear models could also be used for predicting the diameter of pine heartwood at those relative heights. With the help of the pre-dicted diameters a taper curve could be created for the heartwood. The pine heartwood taper curve describes the tapering of the heartwood as function of the tree height. By integrating the taper curve, it was also possible to predict the total volume of the heartwood in a single tree. The models that used tree diameter at breast height and the length of the tree as explanatory variables were able to explain the variation of heartwood diameter on relative heights between 2,5 % and 70 % with coefficient of determination ranging from 0,84 to 0,95 and also recorded a relative RMSE from 15 % to 35 %. Models for relative heights of 85 % and 95 % were not as good as the others (R2-values 0,65 and 0,06 as well as RMSE-values of 74 % and 444 %). Despite not succeeding on all the relative heights, the most important thing is that the models worked best on that area of the stem where most of the heart-wood is located. The volume predictions for single trees based on the heartwood diameter models rec-orded relative RMSE of 35 % and bias of -5 %. Based on the results of the study it shows that exact prediction of pine heartwood diameter is much easier in the base of the stem than in the top part of it. A great deal of variation could be observed whether there was heartwood or not in the top parts of the stem. The volume of heartwood can already be estimated for single trees, but the amount of heartwood can be predicted also in larger scale, such as forest stands. But to get more accurate results in the future, there is a need for more detailed and com-prehensive research data, which would help to determine the still unknown parts of the behaviour of pine heartwood.

The science of fish stock assessment is one that is very resource and labor intensive, with stock assessment models historically being based on data that causes a model to overestimate the strength of a population, sometimes with drastic consequences. The need of cost-effective assessment models and approaches increases, which is why I looked into using Bayesian modeling and networks as an approach not often used in fisheries science. I wanted to determine if it could be used to predict both recruitment and spawning stock biomass of four fish species in the north Atlantic, cod, haddock, pollock and capelin, based on no other evidence other than the recruitment or biomass data of the other species and if these results could be used to lower the uncertanties of fish stock models. I used data available on the RAM legacy database to produce four different models with the statistical software R, based on four different Bayes algorithms found in the R-package bnlearn, two based on continuous data and two based on discrete data. What I found was that there is much potential in the Bayesian approach to stock prediction and forecasting, as our prediction error percentage ranged between 1 and 40 percent. The best predictions were made when the species used as evidence had a high correlation coefficient with the target species, which was the case with cod and haddock biomass, which had a unusually high correlation of 0.96. As such, this approach could be used to make preliminary models of interactions between a high amount of species in a specific area, where there is data abundantly available and these models could be used to lower the uncertanties of the stock assessments. However, more research into the applicability for this approach to other species and areas needs to be conducted.

This study attempts to discover the best predictors of mathematics and language learning outcomes across Kenya, Mozambique, Nigeria, Uganda, and Tanzania by analysing World Bank SDI data and using machine learning methods for variable selection purposes. Firstly, I use the SDI data to show the current fragilities in the quality of education service delivery, while also highlighting deficiencies in student learning outcomes. Then, I use CV Lasso, Adaptive Lasso, and Elastic Net regularisation methods to help discover the best predictors of learning outcomes. While the results from the regularisation methods show that private schools, teacher subject knowledge, and teacher pedagogical skills are good predictors of learning outcomes in a sample combining observations from Kenya, Mozambique, Nigeria, Uganda, and Tanzania, the results fail to infer causality by not distinguishing if unobservable factors are driving the results. To quantify the relationship of key predictors, and for statistical significance testing purposes, I then conduct subsequent OLS analysis. Despite not expecting the true partial derivative effects to be identical to the OLS coefficients presented in this study, this study highlights deficiencies in education service delivery and applies methods which help select key predictors of learning outcomes across the sampled schools in the SDI data.