Prediction with Gaussian processes: from linear regression to linear prediction and beyond

The main aim of this paper is to provide a tutorial on regression with Gaussian processes. We start from Bayesian linear regression, and show how by a change of viewpoint one can see this method as a Gaussian process predictor based on priors over functions, rather than on priors over parameters. This leads in to a more general discussion of Gaussian processes in section 4. Section 5 deals with further issues, including hierarchical modelling and the setting of the parameters that control the Gaussian process, the covariance functions for neural network models and the use of Gaussian processes in classification problems.

Forecasting exchange rates with linear and nonlinear models

In this paper, the exchange rate forecasting performance of neural network models are evaluated against the random walk, autoregressive moving average and generalised autoregressive conditional heteroskedasticity models. There are no guidelines available that can be used to choose the parameters of neural network models and therefore, the parameters are chosen according to what the researcher considers to be the best. Such an approach, however,implies that the risk of making bad decisions is extremely high, which could explain why in many studies, neural network models do not consistently perform better than their time series counterparts. In this paper, through extensive experimentation, the level of subjectivity in building neural network models is considerably reduced and therefore giving them a better chance of Forecasting exchange rates with linear and nonlinear models 415 performing well. The results show that in general, neural network models perform better than the traditionally used time series models in forecasting exchange rates.

A comparison of linear forecasting models and neural networks:an application to Euro inflation and Euro Divisia

Linear models reach their limitations in applications with nonlinearities in the data. In this paper new empirical evidence is provided on the relative Euro inflation forecasting performance of linear and non-linear models. The well established and widely used univariate ARIMA and multivariate VAR models are used as linear forecasting models whereas neural networks (NN) are used as non-linear forecasting models. It is endeavoured to keep the level of subjectivity in the NN building process to a minimum in an attempt to exploit the full potentials of the NN. It is also investigated whether the historically poor performance of the theoretically superior measure of the monetary services flow, Divisia, relative to the traditional Simple Sum measure could be attributed to a certain extent to the evaluation of these indices within a linear framework. Results obtained suggest that non-linear models provide better within-sample and out-of-sample forecasts and linear models are simply a subset of them. The Divisia index also outperforms the Simple Sum index when evaluated in a non-linear framework. © 2005 Taylor & Francis Group Ltd.

An empirical study of salesperson stereotypes amongst UK students and their implications for recruitment

The success of recruitment efforts can have a major impact on sales force effectiveness. Students have long been considered a good source of potential sales recruits, but research has found students have generally negative perceptions of selling as a career. One reason for such perceptions may be negative stereotypes of salespeople held by students. However information on the content of UK sales stereotypes remains anecdotal at best. This study empirically examines UK business students' stereotypes of salespeople using a two-stage approach. Findings suggest that these stereotypes are generally negative. However, we create profiles of salespeople using our findings, and consequently uncover some positive aspects to the stereotype. The study provides instruction on how to use stereotypes in subsequent work, as well as how to utilise the profiles in recruitment efforts.

Perceiving edge blur: linear filtering and a rectifying nonlinearity

We studied the visual mechanisms that encode edge blur in images. Our previous work suggested that the visual system spatially differentiates the luminance profile twice to create the 'signature' of the edge, and then evaluates the spatial scale of this signature profile by applying Gaussian derivative templates of different sizes. The scale of the best-fitting template indicates the blur of the edge. In blur-matching experiments, a staircase procedure was used to adjust the blur of a comparison edge (40% contrast, 0.3 s duration) until it appeared to match the blur of test edges at different contrasts (5% - 40%) and blurs (6 - 32 min of arc). Results showed that lower-contrast edges looked progressively sharper.We also added a linear luminance gradient to blurred test edges. When the added gradient was of opposite polarity to the edge gradient, it made the edge look progressively sharper. Both effects can be explained quantitatively by the action of a half-wave rectifying nonlinearity that sits between the first and second (linear) differentiating stages. This rectifier was introduced to account for a range of other effects on perceived blur (Barbieri-Hesse and Georgeson, 2002 Perception 31 Supplement, 54), but it readily predicts the influence of the negative ramp. The effect of contrast arises because the rectifier has a threshold: it not only suppresses negative values but also small positive values. At low contrasts, more of the gradient profile falls below threshold and its effective spatial scale shrinks in size, leading to perceived sharpening.

Forecasting exchange rates with linear and nonlinear models

In this paper the exchange rate forecasting performance of neural network models are evaluated against random walk and a range of time series models. There are no guidelines available that can be used to choose the parameters of neural network models and therefore the parameters are chosen according to what the researcher considers to be the best. Such an approach, however, implies that the risk of making bad decisions is extremely high which could explain why in many studies neural network models do not consistently perform better than their time series counterparts. In this paper through extensive experimentation the level of subjectivity in building neural network models is considerably reduced and therefore giving them a better chance of performing well. Our results show that in general neural network models perform better than traditionally used time series models in forecasting exchange rates.

Statnote 20: non-linear regression: fitting a general polynomial curve

In some circumstances, there may be no scientific model of the relationship between X and Y that can be specified in advance and indeed the objective of the investigation may be to provide a ‘curve of best fit’ for predictive purposes. In such an example, the fitting of successive polynomials may be the best approach. There are various strategies to decide on the polynomial of best fit depending on the objectives of the investigation.

Data analysis methods in optometry Part 7: multiple linear regression

Multiple regression analysis is a complex statistical method with many potential uses. It has also become one of the most abused of all statistical procedures since anyone with a data base and suitable software can carry it out. An investigator should always have a clear hypothesis in mind before carrying out such a procedure and knowledge of the limitations of each aspect of the analysis. In addition, multiple regression is probably best used in an exploratory context, identifying variables that might profitably be examined by more detailed studies. Where there are many variables potentially influencing Y, they are likely to be intercorrelated and to account for relatively small amounts of the variance. Any analysis in which R squared is less than 50% should be suspect as probably not indicating the presence of significant variables. A further problem relates to sample size. It is often stated that the number of subjects or patients must be at least 5-10 times the number of variables included in the study.5 This advice should be taken only as a rough guide but it does indicate that the variables included should be selected with great care as inclusion of an obviously unimportant variable may have a significant impact on the sample size required.

Data methods in optometry. Part 10: non-linear regression analysis

1. The techniques associated with regression, whether linear or non-linear, are some of the most useful statistical procedures that can be applied in clinical studies in optometry. 2. In some cases, there may be no scientific model of the relationship between X and Y that can be specified in advance and the objective may be to provide a ‘curve of best fit’ for predictive purposes. In such cases, the fitting of a general polynomial type curve may be the best approach. 3. An investigator may have a specific model in mind that relates Y to X and the data may provide a test of this hypothesis. Some of these curves can be reduced to a linear regression by transformation, e.g., the exponential and negative exponential decay curves. 4. In some circumstances, e.g., the asymptotic curve or logistic growth law, a more complex process of curve fitting involving non-linear estimation will be required.

Entry modes and performance of foreign direct investment in China

How does a firm choose a proper model of foreign direct investment (FDI) for entering a foreign market? Which mode of entry performs better? What are the performance implications of joint venture (JV) ownership structure? These important questions face a multinational enterprise (MNE) that decides to enter a foreign market. However, few studies have been conducted on such issues, and no consistent or conclusive findings are generated, especially with respect to China. It’s composed of five chapters, providing corresponding answers to the questions given above. Specifically, Chapter One is an overall introductory chapter. Chapter Two is about the choice of entry mode of FDI in China. Chapter Three examines the relationship between four main entry modes and performance. Chapter Four explores the performance implications of JV ownership structure. Chapter Five is an overall concluding chapter. These empirical studies are based on the most recent and richest data that has never been explored in previous studies. It contains information on 11,765 foreign-invested enterprises in China in seven manufacturing industries in 2000, 10,757 in 1999, and 10,666 in 1998. The four FDI entry modes examined include wholly-owned enterprises (WOEs), equity joint ventures (EJVs), contractual joint ventures (CJVs), and joint stock companies (JSCs). In Chapter Two, a multinominal logit model is established, and techniques of multiple linear regression analysis are employed in Chapter Three and Four. It was found that MNEs, under the conditions of a good investment environment, large capital commitment and small cultural distance, prefer the WOE strategy. If these conditions are not met, the EJV mode would be of greater use. The relative propensity to pursue the CJV mode increases with a good investment environment, small capital commitment, and small cultural distance. JSCs are not favoured by MNEs when the investment environment improves and when affiliates are located in the coastal areas. MNEs have been found to have a greater preference for an EJV as a mode of entry into the Chinese market in all industries. It is also found that in terms of return on assets (ROA) and asset turnover, WOEs perform the best, followed by EJVs, CJVs, and JSCs. Finally, minority-owned EJVs or JSCs are found to outperform their majority-owned counterparts in terms of ROA and asset turnover.

Real options and multinational investments : an empirical study of risk, performance and organisational awareness

DUE TO COPYRIGHT RESTRICTIONS ONLY AVAILABLE FOR CONSULTATION AT ASTON UNIVERSITY LIBRARY AND INFORMATION SERVICES WITH PRIOR ARRANGEMENT This thesis is a cross-disciplinary study of the empirical impact of real options theory in the fields of decision sciences and performance management. Borrowing from the economics, strategy and operations research literature, the research examines the risk and performance implications of real options in firms’ strategic investments and multinational operations. An emphasis is placed on the flexibility potential and competitive advantage of multinational corporations to explore the extent to which real options analysis can be classified as best practice in management research. Using a combination of qualitative and quantitative techniques the evidence suggests that, if real options are explored and exploited appropriately, real options management can result in superior performance for multinational companies. The qualitative findings give an overview of the practical advantages and disadvantages of real options and the statistical results reveal that firms which have developed a high awareness of their real options are, as predicted by the theory, able to reduce their downside risk and increase profits through flexibility, organisational slack and multinationality. Although real options awareness does not systematically guarantee higher returns from operations, supplementary findings indicate that firms with evidence of significant investments in the acquisition of real options knowledge tend to outperform competitors which are unaware of their real options. There are three contributions of this research. First, it extends the real options and capacity planning literature to path-dependent contingent-claims analysis to underline the benefits of average type options in capacity allocation. Second, it is thought to be the first to explicitly examine the performance effects of real options on a sample of firms which have developed partial capabilities in real options analysis suggesting that real options diffusion can be key to value creation. Third, it builds a new decision-aiding framework to facilitate the use of real options in projects appraisal and strategic planning.

The compressive creep and load relaxation properties of a series of high aluminium zinc-based alloys

A new family of commercial zinc alloys designated as ZA8, ZA12, and ZA27 and high damping capacity alloys including Cosmal and Supercosmal and aluminium alloy LM25 were investigated for compressive creep and load relaxation behaviour under a series of temperatures and stresses. A compressive creep machine was designed to test the sand cast hollow cylindrical test specimens of these alloys. For each compressive creep experiment the variation of creep strain was presented in the form of graphs plotted as percentage of creep strain () versus time in seconds (s). In all cases, the curves showed the same general form of the creep curve, i.e. a primary creep stage, followed by a linear steady-state region (secondary creep). In general, it was observed that alloy ZA8 had the least primary creep among the commercial zinc-based alloys and ZA27 the greatest. The extent of primary creep increased with aluminium content to that of ZA27 then declined to Supercosmal. The overall creep strength of ZA27 was generally less than ZA8 and ZA12 but it showed better creep strength than ZA8 and ZA12 at high temperature and high stress. In high damping capacity alloys, Supercosmal had less primary creep and longer secondary creep regions and also had the lowest minimum creep rate among all the tested alloys. LM25 exhibited almost no creep at maximum temperature and stress used in this research work. Total creep elongation was shown to be well correlated using an empirical equation. Stress exponent and activation energies were calculated and found to be consistent with the creep mechanism of dislocation climb. The primary α and β phases in the as-cast structures decomposed to lamellar phases on cooling, with some particulates at dendrite edges and grain boundaries. Further breakdown into particulate bodies occurred during creep testing, and zinc bands developed at the highest test temperature of 160°C. The results of load relaxation testing showed that initially load loss proceeded rapidly and then deminished gradually with time. Load loss increased with temperature and almost all the curves approximated to a logarithmic decay of preload with time. ZA alloys exhibited almost the same load loss at lower temperature, but at 120°C ZA27 improved its relative performance with the passage of time. High damping capacity alloys and LM25 had much better resistance to load loss than ZA alloys and LM25 was found to be the best against load loss among these alloys. A preliminary equation was derived to correlate the retained load with time and temperature.

Theoretical and empirical evaluation of phakic and pseudophakic accomodation

The binding theme of this thesis is the examination of both phakic and pseudophakic accommodation by means of theoretical modelling and the application of a new biometric measuring technique. Anterior Segment Optical Coherence Tomography (AS-OCT) was used to assess phakic accommodative changes in 30 young subjects (19.4 2.0 years; range, 18 to 25 years). A new method of assessing curvature change with this technique was employed with limited success. Changes in axial accommodative spacing, however, proved to be very similar to those of the Scheimpflug-based data. A unique biphasic trend in the position of the posterior crystalline lens surface during accommodation was discovered, which has not been alluded to in the literature. All axial changes with accommodation were statistically significant (p < 0.01) with the exception of corneal thickness (p = 0.81). A two-year follow-up study was undertaken for a cohort of subjects previously implanted with a new accommodating intraocular lens (AIOL) (Lenstec Tetraflex KH3500). All measures of best corrected distance visual acuity (BCDVA; +0.04 0.24 logMAR), distance corrected near visual acuity (DCNVA; +0.61 0.17 logMAR) and contrast sensitivity (+1.35 0.21 log units) were good. The subjective accommodation response quantified with the push-up technique (1.53 0.64 D) and defocus curves (0.77 0.29 D) was greater than the objective stimulus response (0.21 0.19 D). AS-OCT measures with accommodation stimulus revealed a small mean posterior movement of the AIOLs (0.02 0.03 mm for a 4.0 D stimulus); this is contrary to proposed mechanism of the anterior focus-shift principle.

Which is the best solar thermal collection technology for electricity generation in north-west India? Evaluation of options using the analytical hierarchy process

This study of concentrating solar thermal power generation sets out to evaluate the main existing collection technologies using the framework of the Analytical Hierarchy Process (AHP). It encompasses parabolic troughs, heliostat fields, linear Fresnel reflectors, parabolic dishes, compound parabolic concentrators and linear Fresnel lenses. These technologies are compared based on technical, economic and environmental criteria. Within these three categories, numerous sub-criteria are identified; similarly sub-alternatives are considered for each technology. A literature review, thermodynamic calculations and an expert workshop have been used to arrive at quantitative and qualitative assessments. The methodology is applied principally to a case study in Gujarat in north-west India, though case studies based on the Sahara Desert, Southern Spain and California are included for comparison. A sensitivity analysis is carried out for Gujarat. The study concludes that the linear Fresnel lens with a secondary compound parabolic collector, or the parabolic dish reflector, is the preferred technology for north-west India.

Does money matter in inflation forecasting

This paper provides the most fully comprehensive evidence to date on whether or not monetary aggregates are valuable for forecasting US inflation in the early to mid 2000s. We explore a wide range of different definitions of money, including different methods of aggregation and different collections of included monetary assets. In our forecasting experiment we use two non-linear techniques, namely, recurrent neural networks and kernel recursive least squares regression - techniques that are new to macroeconomics. Recurrent neural networks operate with potentially unbounded input memory, while the kernel regression technique is a finite memory predictor. The two methodologies compete to find the best fitting US inflation forecasting models and are then compared to forecasts from a naive random walk model. The best models were non-linear autoregressive models based on kernel methods. Our findings do not provide much support for the usefulness of monetary aggregates in forecasting inflation.