Academic literature on the topic 'Generalized Procrustes Analysis'

The temporal and spatial variability in wind speed and direction was investigated in a study of the mesoscale wind fields in the Canberra region. The statistical description of the winds was based on twelve months of three-hourly data at seventeen sites obtained in a joint program carried out by the Division of Land Use Research, CSIRO, the National Capital Development Commission, and the Bureau of Meteorology. The statistical analysis proceeded in stages. The first two stages were concerned with the determination and examination of averages and measures of dipersion. Information on the temporal variability in regional wind, defined as the average of the winds at the seventeen collection sites, provided the first insight into the important determinants of winds in the region. The data were then categorized on the basis of the information thus obtained, and the averages over time for each site were analysed in each category. The variation between sites revealed the extent of the spatial variability in the region. For each category, for each site, there were perturbations around the average state, and in the last stage of the study, the analysis examined how the perturbations were related across sites using correlation coefficients. Generalized Procrustes Analysis was used, followed by the extensive use of cluster analysis. Linear modelling techniques were used at all stages of the study, not only for wind speed, but also for wind direction which is an angular variate and thus required different modelling procedures. The models related the variables of interest to terrain features such as position, elevation and surface roughness. These models allowed an informed judgement to be made on the likelihood of accurately estimating the winds at other locations in the region using interpolation techniques.

This thesis proposed that theory on human territoriality, particularly the work of Irwin Altman and colleagues, can contribute to a better understanding of the relationship between people and their car. The car has often been labelled a territory, analogous to the home, which plays a significant role in the lives of individuals in motorised societies. A review of the existing literature on both territoriality and driver behaviour suggested that the car conforms to the defining features of a territory, being an important and enduring object that many drivers decorate with personal items and that they protect against invasion by other road users. Despite this literature, there is little empirical research to support the idea that the car is an imp011ant human territory. Using the methodological framework of transactional psychology, this thesis reported on three studies that utilised an array of self-rep01t techniques to answer research questions on the relationship between territoriality and driver behaviour. The research questions enquired as to whether drivers perceive the car and the surrounding road space as territory, whether they see the driving environment as defined in territo1ial terms, and whether these features influence behaviour on the road. In the first study, 13 focus groups were conducted to determine whether drivers' descriptions of the relationship with their car conformed to existing descriptions of territory. Specific interest was taken in whether the car conformed to Brown and Altman's (1981) descriptions of primary (e.g., home), secondary (e.g., workplace) or public (e.g., park) territory. Focus group questions reflected the territorial features of duration, centrality, marking range, marking intention, and response to invasion. Analysis revealed that the car can be considered a territory, with descriptions of the car closely matching Brown and Altman's descriptions of territory types. However, exactly what type of territory the car was depended on the individual. For instance, young drivers were more likely to perceive the car as aligned with notions of primary territory, while drivers of work vehicles were more likely to describe the car as a secondary or even public territory. Fm1her, dimensional variations were present within d1iver groups and within individuals, such that the broad application of the label 'primary', 'secondary' or 'public' territory to desc1ibe the car was not appropriate. As such, it was noted that territorial researchers should determine the nature of a person's relationship with a particu1ar place/object p1ior to making assumptions about what type of territory it is. Study Two examined the constructs (i.e., descriptors) that drivers use to interpret driving situations, hypothesising that they would correspond to territorial themes such as privacy, safety and control. Phase One of the study generated a representative list of 20 common driving situations, through the use of a travel diary, for subsequent use in a repertory grid task. In Phase Two, 101 drivers completed a repertory grid in which they compared the 20 situations on a se1ies of self-generated constructs. Generalised Procrustes analysis of the grids revealed two major dimensions that drivers use to interpret their driving environment, representing the constructs of safety, control, anger, stress and enjoyment. With few exceptions, these constructs were consistently applied by all drivers and across all driving situations. Safety and control are both linked to aspects of human territoriality. Control was also notable for representing both control over whether a situation is entered into, and control over what occurs once in the situation. Although anger, stress and enjoyment are not key territorial themes, this may reflect the comparatively deficient coverage of emotion in territorial literature. In Study Three, the survey responses of 403 drivers were used to test three hypotheses. First, principal components analysis of the relationship with the car revealed that the major themes characterising this association were linked to the dimensions of territoriality desc1ibed by Brown and Altman (1981). In addition, the degree to which each theme was experienced varied as a function of driver age and car ownership, a feature also noted in the focus group study. Second, the majority of drivers were able to nominate a part of the road-space surrounding the car that they saw as part of their territory. The size of this space varied as a function of the driver and the type of car driven. Third, mixed results were found for the effect of territoriality on self-reported aggressive and prosocial driving. Although viewing the car as a defendable space predicted aggressive behaviours, and believing that the car provides autonomy predicted prosocial driving in some situations, other variables directly related to territory were non-significant. As was found in Study Two, the emotional states of enjoyment and anger also played a role in predicting response to the behaviour of others. Together, these studies suggested that the car is a territory, for some drivers analogous to a primary territory such as the home. The manner in which the car, the surrounding road-space, and the driving environment are considered territorial has implications for understanding road user behaviour, including the topics of aggressive and prosocial driving, conformity with road rules, and travel demand management. As such, the findings of this research have potential application in the development of strategies to improve dri ver behaviour, specifically in relation to aggressive and prosocial driving. In addition, the research conducted in this thesis has implications for the further development of theory on human territoriality. Primarily this relates to further elucidating the role of control in territoriality, especially as it differentiates between primary, secondary and public territories. Territoriality theory would also benefit from a more comprehensive understanding of the role that emotion plays in territoriality. Given the exploratory nature of this research, further work is also required to further understand the links between territoriality and driver behaviour.

L'olor sexual és un defecte sensorial que es dóna sobretot en la carn de porcs mascles enters i és deguda principalment a dues substàncies, l'androstenona i l'escatol. Els objectius principals d'aquest estudi són conèixer la distribució d'androstenona i escatol dels porcs del nostre país respecte l'europea, descriure aquests dos compostos sensorialment mitjançant un panel entrenat, estudiar la resposta d'acceptabilitat dels consumidors europeus a la carn de porc mascle enter segons el seu nivell d'androstenona i escatol i veure la influència de la sensibilitat a l'androstenona en la resposta d'acceptabilitat dels consumidors espanyols. Es van medir els nivells d'escatol i androstenona en el greix subcutani de 4.313 mascles enters i 223 femelles, produïts en dues estacions de l'any en 6 països europeus. Per als posteriors estudis sensorials es va seleccionar una submostra de 42 lloms de femelles i 378 de mascles enters, classificats en nou grups segons nivells d'androstenona i escatol alts (>0.99 i >0.21 µg/g respectivament), mitjans o baixos (<0.5 µg/g i <0.10 µg/g). Els degustadors, prèviament seleccionats i entrenats van obtenir el perfil descriptiu pel mètode 'Checklist'. Van trobar per a l'olor els descriptors 'androstenona', 'escatol', 'porc', 'orina', 'suor', 'fems', 'dolç', 'ranci' i 'anormal' i per al flavor els anteriors excepte 'fems'. Van avaluar monàdicament 10 lloms, cuits al forn a 175°C durant 10', en cada una de les 6 sessions d'olor i flavor. Els consumidors (aproximadament 240 per país i rèplica) van valorar 5 mostres segons una escala hedònica de 7 nivells. L'acceptabilitat del flavor de la carn es va avaluar en lloms cuits a 180°C i rescalfats a 250°C i la de l'olor en lloms cuits pel mateix sistema i, a més, escalfats 1 minut en una planxa a 250ºC just abans de servir. En l'assaig espanyol es va fer la prova de la sensibilitat a l'androstenona. La mitjana dels nivells d'androstenona i escatol al nostre país es va situar per sobre de l'europea i va ser de 1.20 µg/g i de 0.19 µg/g respectivament. El percentatge de canals amb nivells d'androstenona superiors a 1.0 µg/g i d'escatol superiors a 0.21 µg/g va ser del 17% al nostre país i del 7% en el global de tots els països participants. En la descripció sensorial dels lloms es va observar una gran confusió dels atributs. Els lloms procedents de mascles enters amb nivells baixos o mitjans d'androstenona i escatol i els lloms procedents de femelles es van caracteritzar sensorialment per l'atribut 'dolç'. La resposta dels consumidors en general va ser sobretot de refús de l'olor de les mostres amb elevats nivells d'escatol, independentment del contingut d'androstenona. Els consumidors espanyols anòsmics o poc sensibles a l'androstenona van donar pitjor nota d'acceptabilitat només als lloms amb elevat nivell d'escatol independentment del nivell d'androstenona. Tanmateix aquest darrer nivell va influir en les notes d'acceptabilitat donades pels consumidors molt sensibles a aquesta substància (24% homes, 37% dones), essent pitjors quant el nivell era elevat o mitjà. Es pot concloure que caldria controlar els nivells d'androstenona i escatol sobretot en països com Espanya en què es consumeixen principalment lloms de mascles enters i es tenen els nivells d'androstenona i escatol elevats. L'olor sexual té una caracterització sensorial complexa, tanmateix, és causa del rebuig o baixa acceptabilitat dels lloms per part dels consumidors, especialment si aquests són sensibles a l'androstenona i, si no es té en compte la sensibilitat, es subestima la importància d'aquesta. Un possible llindar satisfactori per a la majoria de consumidors en les condicions d'aquest estudi seria el refús de canals amb (a) >0.21 µg/g d'escatol, (b) >0.99 µg/g d'androstenona o (c) >0.10 µg/g d'escatol i >0.5 µg/g d'androstenona.
Boar taint is a sensory defect mainly detected in boar meat, mainly due to two substances, androstenone and skatole. The main objectives of this study are to ascertain the androstenone and skatole distribution in the pig population of our country with respect to the group of the other European countries; to obtain the sensory characterisation of both compounds by means of a trained panel; to study the acceptability of the boar meat by the European consumers depending on the levels of androstenone and skatole; and to ascertain the sensitivity to androstenone as an influence on the acceptability of boar to Spanish consumers. The levels of androstenone and skatole in subcutaneous fat were measured in 4313 boars and 223 gilts, produced in 6 European countries during two seasons (replicates). In the sensory studies, 42 loins from gilts and 378 loins from boars were selected and the boars were classified in 9 groups depending on the levels of androstenone and skatole (high:>1.0 and >0.21 µg/g respectively, medium and low: <0.5 µg/g and <0.10 µg/g). Panellists previously selected and trained obtained the profile by the checklist method. 'Androstenone', 'skatole', 'pig', 'urine', 'sweat', 'manure', 'sweet', 'rancid' and 'abnormal' were found as odour attributes. The flavour attributes were the same except for 'manure'. Panellists evaluated in a monadic way 10 loins, cooked at 175ºC in the oven for 10 minutes, in each of the 6 sessions of odour and flavour assessment. Consumers (approximately 240 per country and replicate) evaluated 5 samples in a 7 level hedonic scale. The acceptability of the meat flavour was evaluated in loins cooked at 180ºC and re-cooked at 250ºC and the acceptability of the meat odour was evaluated in loins cooked in the same way and kept for 1 minute on a hot plate at 250ºC just before serving. In our country, consumers' sensibility to androstenone was tested. Mean levels of androstenone and skatole in our country (1.20 µg/g and 0.19 µg/g respectively) were above the European mean. The percentage of carcasses with androstenone and skatole levels higher than 1.0 µg/g and 0.21 µg/g was 17% in our country and 7% in all the European countries studied. In the sensory characterisation of the androstenona and the skatole there was a lot of confusion among attributes.. Boar loins with low and medium levels of androstenone and skatole and gilt loins were sensory characterised by the same attribute 'sweet'. Consumers mainly refused the odour of the loins with high levels of skatole, regardless of the androstenone content. The anosmics or less sensitive to androstenone Spanish consumers gave the worst note of acceptability to the loins with high skatole level, independently of the androstenone level. The androstenone level influence in the notes of acceptability given by the highly sensitive consumers to androstenone (24% men, 37% women), are worse when the level of this substance is high or medium. It can be concluded that the androstenone and skatole levels should be controlled, especially in the countries like Spain, where boar meat is consumed and where there are high levels of androstenone as well as with skatole. Boar taint has a complex sensory characterisation. Loins with boar taint have a low level of acceptability with consumers, especially if they are sensitive to androstenone and, if the sensitivity is not taken into account, the importance of androstenone is underestimated. A possible threshold that satisfies the most important part of the consumers in the conditions of this study would be the rejection of carcasses with (a) >0.21 µg/g of skatole, (b) >0.99 µg/g of androstenone or (c) >0.10 µg/g of skatole and >0.5 µg/g of androstenone.

One of the most vital duties for engineers is to preserve life and nature by utilising safe designs that take into account environmental standards and monitoring the performance of structures against design criteria. Furthermore, monitoring can be used to determine any required maintenance of an important structure following a catastrophic event. Numerous different techniques and instruments can be employed for such a purpose with different requirements producing different results. For instance, some techniques need to embed sensors inside the building, such as Geotechnical Sensors. Others can offer high quality, but with a low point density and require fixed stations and targets, like Total Stations (TS). In such cases, the location of deformation tends to be known, such as in dams, bridges, and high-rise buildings. However, this is not always the case where it might be hard to expect deformation location as in the case of historic ruins where each part of the structure could be subject to deformation. The challenge in such case is to detect the deformation without any previous knowledge. Remote Sensing (RS) techniques, such as Digital Photogrammetry, Synthetic Aperture Radar (SAR), Interferometric Synthetic Aperture Radar (InSAR), and Terrestrial Laser Scanner (TLS) can be solutions for such an issue. Interestingly, many researchers are focusing on using TLS for monitoring owing to the great spatial resolution system can offer. However, there are three challenges in using TLS in monitoring: the first one is a huge amount of data and the difficulty of handling it; the second one is the difficulty of comparing between two epochs because observations of TLS are not repeatable; and the third issue is the noise which is attached to the data. The first problem is solved by segmentation and point structure while the second and the third ones still need more investigation, although some interesting researches have been done in this area. The aim of this research is to develop a new approach to detect and localise unpredictable deformation. It is based on TLS measurements and Generalised Procrustes Analysis (GPA) techniques to determine deformation vectors, while boxing structure and F-test are used to detect and localise deformation. In summary, after applying this approach, the whole concerned building is represented as parts, for each of which the displacement vector and the deformation probability are estimated. Ultimately, it is possible to monitor any part through different epochs. In addition, through this technique, it is possible to determine deformations - not just between two epochs, but for sequences of them. This can give more reliable results. Four validation experiments have been conducted. The first test was designed to assess the performance of the developed software and to fix some variables. Therefore, it was based on simulated data with controlled white noise, distributed according to the normal distribution, and simulated deformations. The results of this test revealed the success of the proposed algorithm to detect and to localise deformations. In addition, it showed the success when no deformations exist. Furthermore, optimistically, it could observe deformations with magnitude less than the noise level; however, the probability was only 40%. Correspondingly, real scan data with simulated deformations was used in the second test. The purpose of this test is to examine the performance of the proposed method in case of real errors budget. However, the short range of the test (about 10m), a featureless scanned area (wall only), and scanning from one position for all epochs (no need for registration) can reduce errors to a minimum. Results of this test showed the success of the proposed method to detect and localise deformations. Potentially, it can give indications for areas with deformations less than the noise level. Furthermore, results of the proposed method can be considered better than that of CloudCompare software. The third test was conducted to examine the performance of the proposed technique regarding different materials and textures. For this purpose, the Nottingham Geospatial Building (NGB) was selected with more extensive ranges (between 20-25 m). Similar to the second test, all measurements were taken from the same scanner position. To some extent, the proposed technique succeeded to detect and to localise deformations. However, the researcher does not recommend it for monitoring modern and complicated buildings, instead it has been developed for monitoring historic ruins. Finally, the proposed method was applied on the Bellmanpark Limekiln, Clitheroe, Lancashire monitoring project. This is a live project for Historic England and addresses a historic building that currently has some structural issues. The outcome of the proposed method revealed deformations in the faces South East (SE) and North East (NE). From examining these faces, three deformed areas were found: two in the face SE and one in the face NE, which might cause some cracks appeared in these faces. Alternatively, the CloudCompare software has been employed to detect deformation. Although results coincide with the proposed method for detected deformations, it cannot locate these deformations very well because it diffused over a wide area. In addition, it cannot determine actual directions of the deformations unlike the proposed method.

Geralmente, nos experimentos genótipo por ambiente (G × E) é comum observar o comportamento dos genótipos em relação a distintos atributos nos ambientes considerados. A análise deste tipo de experimentos tem sido abordada amplamente para o caso de um único atributo. Nesta tese são apresentadas algumas alternativas de análise considerando genótipos, ambientes e atributos simultaneamente. A primeira, é baseada no método de mistura de máxima verossimilhança de agrupamento - Mixclus e a análise de componentes principais de 3 modos - 3MPCA, que permitem a análise de tabelas de tripla entrada, estes dois métodos têm sido muito usados na área da psicologia e da química, mas pouco na agricultura. A segunda, é uma metodologia que combina, o modelo de efeitos aditivos com interação multiplicativa - AMMI, modelo eficiente para a análise de experimentos (G × E) com um atributo e a análise de procrustes generalizada, que permite comparar configurações de pontos e proporcionar uma medida numérica de quanto elas diferem. Finalmente, é apresentada uma alternativa para realizar imputação de dados nos experimentos (G × E), pois, uma situação muito frequente nestes experimentos, é a presença de dados faltantes. Conclui-se que as metodologias propostas constituem ferramentas úteis para a análise de experimentos (G × E) multiatributo.
Usually, in the experiments genotype by environment (G×E) it is common to observe the behaviour of genotypes in relation to different attributes in the environments considered. The analysis of such experiments have been widely discussed for the case of a single attribute. This thesis presents some alternatives of analysis, considering genotypes, environments and attributes simultaneously. The first, is based on the mixture maximum likelihood method - Mixclus and the three-mode principal component analysis, these two methods have been very used in the psychology and chemistry, but little in agriculture. The second, is a methodology that combines the additive main effects and multiplicative interaction models - AMMI, efficient model for the analysis of experiments (G×E) with one attribute, and the generalised procrustes analysis, which allows compare configurations of points and provide a numerical measure of how much they differ. Finally, an alternative to perform data imputation in the experiments (G×E) is presented, because, a very frequent situation in these experiments, is the presence of missing values. It is concluded that the proposed methodologies are useful tools for the analysis of experiments (G×E) multi-attribute.

碩士
國立成功大學
工業設計學系碩博士班
98
The purpose of this study is to verify if General Procrustes Analysis (GPA) will increase the precision of the results of the sample cluster. This study uses the front view outline of each car (total 50 cars) as a sample, and we classify the sample into nine parts. We use RHINO to draw the outline and find out the coordinate of the points. After adjusting the points using GPA, we use cluster analysis to group our samples, In the past, coordinate of the points is usually classfied through cluster anaalysis directly. However, the clustering results were not always satisfied. In this study, we try to adjust the data through GPA before applying the cluster analysis. The study also tries to adopt Support Vector Regression (SVR) method. We input the clustering results and emotional vocabulary to train the emotional model. The purpose is to verify if there are any differences between the GPA adjusted data and non-adjusted ones. The results suggest that the clustered data after using GPA indeed show lower RMSE. The trained system could be adopted as an expert system to predict the emotional vocabulary scores of new cars.

碩士
國立成功大學
工業設計學系碩博士班
98
This study focuses on applying Generalized Procrustes Analysis (GPA) in conducting Kansei Engineering System (KES) research on flash drives, which is hoped to introduce “flexible questionnaire assessments” as a way of evaluating product perception through the conducting of methodology. And by using of the method, it would become not only a supportive technique of the traditional fashion, but also the primarily measurements of how consumers are thinking. The study includes three aspects for comparing the differences between traditional methods and the proposed method: The extraction of individual features, sample preference mapping, and the effectiveness of clustering. Further discussion would be on the application of subjects’ responsive features inside of the KES research. The result shows that 1. Using GPA on normal questionnaires can derive results similar to the traditional analytical method. 2. GPA can even extract constructs out of flexible questionnaires for further preference mapping of the samples, which normal methods cannot. 3. Conducting flexible questionnaires assessments would have a more comprehensive feedback on sample evaluations than the normal questionnaires. 4. With the use of GPA, the subjects’ affective response can be retrieved and applied, which raises the effectiveness of clustering analysis.

The measurement and understanding of user emotions elicited by product appearance are critical elements of the product development process and have been interesting design challenges for many years. This paper proposes an original emotion measurement method, called Auditory Parameter Method. It is a non-verbal technique, which uses sounds and association tests for evaluating a set of products (given by their pictures). It provides an assessment of these products according to a series of emotional dimensions. We present a methodological framework to build the links between user’s emotional responses and geometrical features of product, by using a glasses frame 3D model as application case. Analysis of Variance techniques are employed to examine how various shape factors influence users’ emotional responses to 3D model. To demonstrate the effectiveness of our protocol, we compare the proposed method with the conventional Semantic Differential using Principal Component Analysis and Generalized Procrustes Analysis. The new protocol demonstrates interesting qualities to collect the intuitive emotions of user.