Statistical Modeling

This paper introduces and defends an account of model-based science that I dub model pluralism. I argue that despite a growing awareness in the philosophy of science literature of the multiplicity, diversity, and richness of models and modeling-practices, more radical conclusions follow from this recognition than have previously been inferred. Going against the tendency within the literature to generalize from single models, I explicate and defend the following two core theses: (i) any successful analysis of models must target sets of models, their multiplicity of functions within science, and their scientific context and history and (ii) for almost any aspect x of phenomenon y, scientists require multiple models to achieve scientific goal z.

Although lovastatin production has been reported for different microorganism species, there is limited information about lovastatin production by basidiomycetes. The optimization of culture parameters that enhances lovastatin production by Omphalotus olearius OBCC 2002 was investigated, using statistically based experimental designs under solid state fermentation. The Plackett Burman design was used in the first step to test the relative importance of the variables affecting production of lovastatin. Amount and particle size of barley were identified as efficient variables. In the latter step, the interactive effects of selected efficient variables were studied with a full factorial design. A maximum lovastatin yield of 139.47mg/g substrate was achieved by the fermentation of 5g of barley, 1-2mm particle diam., at 28°C. This study showed that O. olearius OBCC 2002 has a high capacity for lovastatin production which could be enhanced by using solid state fermentation with novel and cost-effective substrates, such as barley.

Introduction. Agroforests are often surrogate habitats for threatened bryophytes and other organisms. This is important in urban settings where natural forests are often absent.
Unfortunately, the benefits of urban agroforests for biodiversity are rarely studied. Methods. To better understand the bryophyte diversity of agroforests, we studied corticolous bryophytes (≤ 2 m) on 30 trees each in Citrus L. and Mangifera L. plantations in Nigeria.
Key results. Nineteen species were recorded: 13 leafy liverworts and six mosses. Twelve species were recorded per plantation. Five species were shared between the two agroforests, indicating that a diversity of agroforests is ideal for preserving bryophyte diversity. Bryophyte bundance in the agroforests was similar, but the species were more evenly distributed and the diversity was higher in the mango agroforest. The frequent pruning of the crown (a standard horticultural practice) in the mango agroforest may relate to these differences. Calymperes
palisotii was the most abundant species recorded in the study and was particularly abundant in the Citrus plantation. Generally, corticolous species were absent from the tree base, but occur at a higher mean height on mango (1.39 m) than on Citrus (1.11 m).
This may be due to differences in species, or management practices. The height range for epiphytic bryophytes is slightly broader (0.48–1.80 m) in the Citrus plantation than in the mango plantation (1.17–1.60 m). However, the height of occurrence of bryophyte species in Citrus was generally lower, except for the moss Rhacopilopsis trinitensis, which has the maximum mean height on the trunk.

Several methods which have been adopted to analyze multi-category data yields unsatisfactory results because of strict assumptions regarding normality, linearity, and homoscedasticity. As a result, Multinomial logistic regression is considered as an alternative because it does not assume normality, linearity, or homoscedasticity (Hosmer & Lemeshow, (2000)). The study attempted to use Maximum likelihood estimation and predicted probability to model Maternal Health Care Services data based on a set of explanatory variables. Also to determine the indices that affect Mortality rate. The result shows that wealth index has a significant impact on the use of public and private health delivery facilities. Educational level, antenatal care, assistance during delivery and place of residence are also important factors in assessing Maternal Health Care Services. Finally, the study revealed that educated women, who are wealthy, living in urban areas and who received antenatal care services and assistance during delivery are more likely to utilize Maternal Health Care Services (MHCS)

Background: With a maternal mortality rate of 576 deaths per 100,000 live births, Nigeria accounts for about 10% of all maternal deaths, globally, and has the second highest mortality rate in the world. This high mortality rate makes maternal health a huge public health problem in the country. This paper, therefore, aimed at investigating socio-demographic factors affecting the utilization of Maternal Health Care (MHC) in the context of hierarchical modelling. Methods: Data were extracted from the Nigerian Demographic and Health Survey, 2013. The data have a hierarchical structure, with the 20,116 Ever-Married Women nested within their respective states of residence. Three different hierarchical logistic regression models were formulated to allow for comparison of outcomes between clusters. Findings: A proportion of opposed odds ratio of 0.45 indicate that in 45% of pair-wise comparisons between the urban and rural residence, the odds of MHC utilization was higher at an urban residence than at a rural residence by 1.388 times. The Median Odds Ratio (MOR) for Model 2 indicated the odds of MHC utilization was less than 2.41 for a woman in a state at higher risk compared to a different woman in a state at lower risk. The intra-class correlation coefficient revealed a 40% (Model 1), 21% (Model 2) and 16% (Model 3) chances of utilizing MHC, explained by between-states differences, respectively. Conclusion: In order to close the variation in healthcare delivery in Nigeria, there is a need for government to execute state-specific interventions that would allow fair distribution and utilization of MHC.

Frequency and duration of floods are analyzed using the global flood database of the Dartmouth Flood Observatory (DFO) to explore evidence of trends during 1985-2015 at global and latitudinal scales. Three classes of flood duration (i.e., short: 1-7, moderate: 8-20, and long: 21 days and above) are also considered for this analysis. The nonparametric Mann-Kendall trend analysis is used to evaluate three hypotheses addressing potential monotonic trends in the frequency of flood, moments of duration, and frequency of specific flood duration types. We also evaluated if trends could be related to large-scale atmospheric teleconnections using a generalized linear model framework. Results show that flood frequency and the tails of the flood duration (long duration) have increased at both the global and the latitudinal scales. In the tropics, floods have increased 4-fold since the 2000s. This increase is 2.5-fold in the north midlatitudes. However, much of the trend in frequency and duration of the floods can be placed within the long-term climate variability context since the Atlantic Multidecadal Oscillation, North Atlantic Oscillation, and Pacific Decadal Oscillation were the main atmospheric teleconnections explaining this trend. There is no monotonic trend in the frequency of short-duration floods across all the global and latitudinal scales. There is a significant increasing trend in the annual median of flood durations globally and each latitudinal belt, and this trend is not related to these teleconnections. While the DFO data come with a certain level of epistemic uncertainty due to imprecision in the estimation of floods, overall, the analysis provides insights for understanding the frequency and persistence in hydrologic extremes and how they relate to changes in the climate, organization of global and local dynamical systems, and country-scale socioeconomic factors.

C and N isotopes (collagen), trace elements (apatite), and WARN modeling conducted on sub-adults from prehistoric İkiztepe.
•The weaning age is suggested enabling greater understanding of the weaning process and population dynamics.
•WARN has emphasised the lag effect in living bone as something to be considered when examining the weaning process.
•Reconsideration of the utility of trace element analyses, and the importance of utilising a holistic approach.

The main purpose of this study is to comparatively assess the susceptibility of earthquake-triggered landslides in the island of Lefkada (Ionian Islands, Greece) using two different statistical analysis models, a bivariate model represented by frequency ratio (FR), and a multivariate model represented by logistic regression (LR). For the implementation of the models, the relationship between geo-environmental factors contributing to landslides and documented events related to the 17th November 2015 earthquake was investigated by geographic information systems (GIS)-based analysis. A landslide inventory with events attributed to the specific earthquake was prepared using satellite imagery interpretation and field surveys. Eight factors: Elevation, slope angle, slope aspect, distance to main road network, distance to faults, land cover, geology, and peak ground acceleration (PGA), were considered and used as thematic data layers. The prediction capability of the models and the accuracy of the resulting susceptibility maps were tested by a standard validation method, the receiver operator characteristic (ROC) analysis. Based on the validation results, the output map with the highest reliability could potentially constitute an ideal basis for use within regional spatial planning as well as for the organization of emergency actions by local authorities.

Music is historically the most widely popular and accessible artistic form expression. Its influence goes beyond comprehension and influences the lives of millions everyday. While music dates back as far as the modern human, it has also had mass effect on society in the last 10 years. This yields it vital for us to understand ongoing trends in its characteristics to better understand its future significance. Some significant research has been conducted on the topic in the past, such as Shane Snow’s analysis of key verbiage and content in songs dating back to 1965. While substantial, many of these works cast a net over only specific qualitative or quantitative characteristics and trends of musical pieces. This gap in research yields our team to hypothesize the value of more in-depth multi-variate analyses - possibly concluding temporal correlations that we can use to predict the music of the future. Using a large song collection evenly distributed over time period and artist familiarity and normalized for characteristic 'extremity', our research aims utilize the SpotipyAPI to prove a temporal analysis of the dataset's quantifiable characteristics.

In this paper, optimal experimental design is applied for the characterization of the microbial inactivation of Listeria monocytogenes under non-isothermal conditions, modelled by the Bigelow model. These conditions simulate industrial processes, where different temperature profiles can be applied for food processing. Here, the aim was to find the best time nodes for measurements given a fixed number of observations and a final time process. Results show that, compared to a “classic” uniform time distribution for the observations, the optimal distribution of observation times results in a more accurate description of the response of the microbial population. The OED resulted in a reduction in the relative standard errors for the estimated D and z-values of 25% and 80%, respectively.
Prediction intervals of the microbial counts were calculated based on the parameters estimated with both designs (i.e., uniform and optimal design). The prediction interval generated using the parameters estimated from the optimal design is notably narrower than the one obtained when the “uniform” experiment is considered, thus providing a more accurate description of the thermal resistance of the microorganism. Although the Bigelow model has been used in this particular case, the functions developed have been wrapped in an R package (bioOED), which is freely available and can be used for any other type of microorganism and/or inactivation model.

Beginning in the early 1920s, Bolshevik leaders proclaimed the need to radically revise the pre-revolutionary legacy of children’s literature and to create new Soviet books for children. In our
paper, we seek to disentangle what factors played a role in the chances of legacy authors and works to be included in the limited selection of appropriated children’s classics by the 1930s.
Based on thе comprehensive bibliographic data on books for children printed between 1918 and 1932 along with several authoritative Soviet sources recommending books for children, we use statistical modeling to assess what authorities effectively served as a kind of “security certificate” protecting certain authors and books from the default purge policy. Our results indicate that inclusion in the 1923 Narkompros list of authors whose work was pronounced a state monopoly, as well as inclusion in the Gorky’s list of books suggested for his “World
Literature” publishing house both had a significant positive  effect on the number of printings by the given author. Contrary to our expectations, the popularity of the author in the pre-revolutionary anthologies for children did not promise any significant publishing growth prospects in the 1920s and early 1930s.

Dès le début des années 1920, les dirigeants bolchéviques ont proclamé la nécessité de réviser
radicalement l’héritage prérévolutionnaire de la littérature jeunesse et de créer de nouveaux
livres soviétiques pour enfants. Dans notre article, nous cherchons à dégager les facteurs qui ont
favorisé les chances des auteurs et des œuvres patrimoniales d’être inclus dans la courte liste des
classiques pour enfants, dans les années 1930. Sur la base de données bibliographiques complètes
sur les livres pour enfants imprimés entre 1918 et 1932, et de plusieurs sources soviétiques
faisant autorité pour recommander les livres pour enfants, nous effectuons une modélisation
statistique pour évaluer quelles autorités ont effectivement servi de « sauf-conduit » protégeant
certains auteurs et livres contre la politique de purge par élimination. Nos résultats indiquent
que l’inclusion dans la liste du Narkompros de 1923 des auteurs dont l’œuvre a été déclarée
monopole d’État, ainsi que l’inclusion dans la liste des livres proposés par Gorki pour sa maison
d’édition « Littérature du monde » ont eu un effet positif significatif sur le nombre d’éditions de
cet auteur. Contrairement à nos attentes, la popularité d’un auteur dans les anthologies pour
enfants d’avant la révolution n’est pas corollée à une perspective de croissance significative de sa
réédition dans les années 1920 et au début des années 1930.

Начиная с начала 1920-х годов большевистские лидеры провозгласили необходимость радикального пересмотра революционного наследия детской литературы и создания новых советских книг для детей. В нашей работе мы стремимся разобраться, какие факторы повлияли на шансы авторов и их произведений быть включенными в ограниченный набор признанных детских классиков к 1930-м годам. На основе библиографических данных о книгах для детей, напечатанных в период с 1918 по 1932 год, а также ряда авторитетных советских источников, рекомендующих книги для детей, мы используем статистическое моделирование для оценки того, какие органы власти эффективно служили своего рода «охранной грамотой», защищающей определенных авторов и книги от того,
чтобы быть  исключенными из легитимного наследия.
Наши результаты свидетельствуют о том, что включение в список Наркомпроса 1923 года авторов, чьи
произведения были объявлены государственной собственностью, а также включение
в список книг Максима Горького, предложенных им для издательства «Всемирная литература», оказали значительное положительное влияние на количество печатных
изданий данного автора. Вопреки нашим ожиданиям, популярность автора в
дореволюционных антологиях для детей не обещала сколько-нибудь значительных
перспектив издательского роста в 1920-е и начале 1930-х годов.

This study resumes our research into variations in settlement patterns from the Neolithic to the Migration Period (5600 BC–570 AD). After using a large dataset of less precisely localized finds from the Czech Republic, we now examine data from large-scale surface surveys. The higher spatial precision allows us to analyse settlement activities in terms of quantity, size, duration, continuity, stability and degree of complexity. First, we analysed the data using descriptive methods regarding its spatial structure, dating and environmental setting. We assessed the possibility of integrating data from surface and subsurface research. In the second stage, we reconstructed possible configurations of habitation areas and their adjacent primary production areas (settlement cores) and chronologically ordered the finds using algorithmic modelling. A more detailed phasing transcending the chronological resolution of the data was achieved by using mutual spatial exclusion of settlement cores as a chronological marker. The resulting ordering was then analysed using probabilistic methods. The results portray the intensity of settlement activities during various periods as well as changes in their structural organization. The observed patterns suggest higher-order social organization starting in the Early Bronze Age, culminating in the Final Bronze Age followed by a gradual decline. In later periods, we observe hot spots in the landscape with stable habitation over hundreds of years. The method used is widely applicable for all periods of agricultural prehistory regardless of region. Original data and an example implementation of the method are available as supplementary material and online at https://github.com/demjanp/chrono_spatial_modelling.

The paper focuses on the geostatistical analysis of the data set on the Philippine archipelago. The research question is understanding variability in several geospatial parameters (geology, geomorphology, tectonics and bathymetry) in different segments of the study area. The initial data set was generated in QGIS by digitizing 25 cross-sectioning profiles. The data set contained information on the geospatial parameters in the samples by profiles. Modelling and statistical analysis were performed in SPSS IBM Statistics software. The analysis of the topography shows strong variability of the elevations in the samples with the extreme depths in the central part of the study area (profile 13 with-9,400 m) and highest elevations in its southwestern part (profile 17 with 1950 m). The analysis of the geological classes and lithology shows maximal samples of the basic volcanic rocks (40,40%) followed by mixed sedimentary consolidated rocks (31,90 %). Pairwise analysis of the sediment thickness and slope aspect demonstrates correlation between these two variables with the maximal sediment layer in the profiles 1-4 crossing the Philippines. The hierarchical dendrogram clustering of the bathymetry by three approaches shown maximal correlation of 5 clusters containing profile groups: 12-18 (centre), 22-25 (south-west), 1-2 (north), 7-8 (north-east), 19-21 (south-west). Other profiles show lesser similarities in the bathymetric patterns. The forecasting models were computed for the geospatial variables showing gradual increase in the gradient angles southwards and increased values for the sediment thickness in the north. Technically, the results proved effectiveness of the SPSS application of the geological data modelling.

The decision to specify model predictors as fixed or random effects is not always clear cut and at times different interpretations regarding their nature might be possible. This study investigated the imputation of missing counts of powerboat retrievals from camera data using climatic and temporal variables as predictors. The temporal variables could be treated as fixed or random effects. To evaluate the impact of the treatment of the temporal model predictors, patterns of observed outages were applied to a set of complete 12-month hourly camera data. The proportion of missing data ranged from 0.06 to 0.31. A variety of generalized linear and mixed models built on the Bayesian sequential regression multiple imputation framework were formulated to impute the missing values. The models were assessed using the percentage bias, root-mean-square error and skill score. Results from ten replicated multiple imputation scheme showed that the mixed effect models obtained plausible mean estimates of the total number of powerboat retrievals with less variability than those from fixed effect models. A comparison with predictive mean matching was also performed which showed that the popular predictive mean matching performed worse.