Sara Sjöstedt de Luna

The Bootstrap and Kriging Prediction Intervals

Kriging is a method for spatial prediction that, given observations of a spatial process, gives the optimal linear predictor of the process at a new specified point. The kriging predictor may be used to define a prediction interval for the value of interest. The coverage of the prediction interval will, however, equal the nominal desired coverage only if it is constructed using the correct underlying covariance structure of the process. If this is unknown, it must be estimated from the data. We study the effect on the coverage accuracy of the prediction interval of substituting the true covariance parameters by estimators, and the effect of bootstrap calibration of coverage properties of the resulting ‘plugin’ interval. We demonstrate that plugin and bootstrap calibrated intervals are asymptotically accurate in some generality and that bootstrap calibration appears to have a significant effect in improving the rate of convergence of coverage error.

How will Your Workload Look Like in 6 Years? Analyzing Wikimedia's Workload

Accurate understanding of workloads is key to efficient cloud resource management as well as to the design of large-scale applications. We analyze and model the workload of Wikipedia, one of the worlds largest web sites. With descriptive statistics, time-series analysis, and polynomial splines, we study the trend and seasonality of the workload, its evolution over the years, and also investigate patterns in page popularity. Our results indicate that the workload is highly predictable with a strong seasonality. Our short term prediction algorithm is able to predict the workload with a Mean Absolute Percentage Error of around 2%.

Subsampling Methods to Estimate the Variance of Sample Means Based on Nonstationary Spatial Data With Varying Expected Values

Subsampling and block resampling methods have been suggested in the literature to nonparametrically estimate the variance of statistics computed from spatial data. Usually stationary data are required. However, in empirical applications, the assumption of stationarity often must be rejected. This article proposes nonparametric methods to estimate the variance of (functions of) sample means based on nonstationary spatial data using subsampling. We assume that data are observed on a lattice in some region of R2. In the data that we consider, the information in the different picture elements (pixels) of the lattice are allowed to come from different distributions, with smoothly varying expected values, or with expected values decomposed additively into directional components. Furthermore, pixels are assumed to be locally dependent, and the dependence structure is allowed to differ over the lattice. Consistent variance estimators for (functions of) sample means, together with convergence rates in mean square, are provided under these assumptions. An example with applications to forestry, using satellite data, is discussed.

A Method to Estimate Fibre Length Distribution in Conifers Based on Wood Samples from Increment Cores

Summary We propose a method to estimate fibre length distribution in conifers based on wood samples from increment cores processed by automatic optical fibre-analysers. Automatic fibre-analysers are unable to distinguish: a) fibres from other tissues, “fines”, and b) cut from uncut fibres. However, our proposed method can handle these problems if the type of distributions that fibre lengths and fines follow is known. In our study the length distributions of fines and fibres were assumed to follow truncated normal distributions, characterised by means and standard deviations of the two distributions. Parameter estimates were obtained by the maximum likelihood method. Wood samples from two 22-year-old Scots pine trees at breast height were used to evaluate the performance of the method. From stem discs at 1.5 m, adjacent samples of 5 mm increment cores and wood pieces were taken. The cores were trimmed 1 mm at each side and samples were, after maceration, analysed in a Kajaani FiberLab 3.0. The results showed that the method works well and gives a possibility to distinguish fine and fibre length distribution.

Forecasting risk of tick-borne encephalitis (TBE): Using data from wildlife and climate to predict next year's number of human victims

Abstract Background: Over the past quarter century, the incidence of tick-borne encephalitis (TBE) has increased in most European nations. However, the number of humans stricken by the disease varies from year to year. A method for predicting major increases and decreases is needed. Methods: We assembled a 25-y database (1984–2008) of the number of human TBE victims and wildlife and climate data for the Stockholm region of Sweden, and used it to create easy-to-use mathematical models that predict increases and decreases in the number of humans stricken by TBE. Results: Our best model, which uses December precipitation and mink (Neovison vison, formerly Mustela vison) bagging figures, successfully predicted every major increase or decrease in TBE during the past quarter century, with a minimum of false alarms. However, this model was not efficient in predicting small increases and decreases. Conclusions: Predictions from our models can be used to determine when preventive and adaptive programmes should be implemented. For example, in years when the frequency of TBE in humans is predicted to be high, vector control could be intensified where infested ticks have a higher probability of encountering humans, such as at playgrounds, bathing lakes, barbecue areas and camping facilities. Because our models use only wildlife and climate data, they can be used even when the human population is vaccinated. Another advantage is that because our models employ data from previously-established databases, no additional funding for surveillance is required.

Red fox and tick-borne encephalitis (TBE) in humans: Can predators influence public health?

Analysing datasets from hunting statistics and human cases of tick-borne encephalitis (TBE), we found a positive correlation between the number of human TBE cases and the number of red fox (Vulpes vulpes). Time lags were also present, indicating that high numbers of red fox in 1 y translated into high numbers of human TBE cases the following y. Results for smaller predators were mixed and inconsistent. Hares and grouse showed negative correlations with human TBE cases, suggesting that they might function as dilution hosts. Combining our findings with food web dynamics, we hypothesize a diversity of possible interactions between predators and human disease – some predators suppressing a given disease, others enhancing its spread, and still others having no effect at all. Larger-sized predators that suppress red fox numbers and activity (i.e. wolf, Canis lupus; European lynx, Lynx lynx) were once abundant in our study area but have been reduced or extirpated from most parts of it by humans. We ask what would happen to red foxes and TBE rates in humans if these larger predators were restored to their former abundances.

Measuring Cloud Workload Burstiness

Workload burstiness and spikes are among the main reasons for service disruptions and decrease in the Quality-of-Service (QoS) of online services. They are hurdles that complicate autonomic resource management of datacenters. In this paper, we review the state-of-the-art in online identification of workload spikes and quantifying burstiness. The applicability of some of the proposed techniques is examined for Cloud systems where various workloads are co-hosted on the same platform. We discuss Sample Entropy (Samp En), a measure used in biomedical signal analysis, as a potential measure for burstiness. A modification to the original measure is introduced to make it more suitable for Cloud workloads.

Adjusting for fibre length-biased sampling probability using increment cores from standing trees

In a previous article (Mörling et al. 2003), we described a method to estimate the true fibre length distribution of conifers based on wood samples from 5-mm increment cores. However, the length bias problem arising from the fact that longer cells are more likely to be sampled in an increment core was neglected. The calculated true celllength distribution in the standing tree described in the quoted paper corresponded to the cell length distribution of those cells that at least partially appear in the increment core. This led to underestimation of the proportion of short fibres and overestimation of the proportion of long fibres. In the present study we demonstrate how to correct for this length bias. We used the same data set as presented previously to demonstrate the effect of the refined method.

Clustering misaligned dependent curves applied to varved lake sediment for climate reconstruction

In this paper we introduce a novel functional clustering method, the Bagging Voronoi K-Medoid Aligment (BVKMA) algorithm, which simultaneously clusters and aligns spatially dependent curves. It is a nonparametric statistical method that does not rely on distributional or dependency structure assumptions. The method is motivated by and applied to varved (annually laminated) sediment data from lake Kassjön in northern Sweden, aiming to infer on past environmental and climate changes. The resulting clusters and their time dynamics show great potential for seasonal climate interpretation, in particular for winter climate changes.

Method for accurate fiber length determination from increment cores for large-scale population analyses in Norway spruce

Abstract Fiber (tracheid) length is an important trait targeted for genetic and silvicultural improvement. Such studies require large-scale non-destructive sampling, and accurate length determination. The standard procedure for non-destructive sampling is to collect increment cores, singularize their cells by maceration, measure them with optical analyzer and apply various corrections to suppress influence of non-fiber particles and cut fibers, as fibers are cut by the corer. The recently developed expectation-maximization method (EM) not only addresses the problem of non-fibers and cut fibers, but also corrects for the sampling bias. Here, the performance of the EM method has been evaluated by comparing it with length-weighing and squared length-weighing, both implemented in fiber analyzers, and with microscopy data for intact fibers, corrected for sampling bias, as the reference. This was done for 12-mm increment cores from 16 Norway spruce (Picea abies (L.) Karst) trees on fibers from rings 8–11 (counted from pith), representing juvenile wood of interest in breeding programs. The EM-estimates provided mean-fiber-lengths with bias of only +2.7% and low scatter. Length-weighing and length2-weighing gave biases of -7.3% and +9.3%, respectively, and larger scatter. The suggested EM approach constitutes a more accurate non-destructive method for fiber length (FL) determination, expected to be applicable also to other conifers.