Artificial neural networks and geostatistical models for housing valuations in urban residential areas

Property valuation studies often use classical statistics techniques. Among these techniques, the Artificial Neural Networks are the most applied, overcoming the inflexibility and the linearity of the hedonic models. Other researchers have used Geostatistics techniques, specifically the Kriging Method, for interpreting spatial-temporal variability and to predict housing unit prices. The innovation of this study is to highlight how the Kriging Method can help to better understand the urban environment, improving the results obtained by classical statistics. This study presents two different methods that share the general objective of extracting information regarding a city’s housing from datasets. The procedures applied are Ordinary Kriging (Geostatistics) and Multi-Layer Perceptron algorithm (Artificial Neural Networks). These methods were used to predict housing unit prices in the municipality of Pozuelo de Alarcon (Madrid). The implementation of both methods provides us with the urban characteristics of the study area and the most significant variables related to price. The main conclusion is that the Ordinary Kriging models and the Neural Networks models, applied to predicting housing unit prices are necessary methodologies to improve the information obtained in classical statistical techniques.

Abbreviations: ANN: Artificial Neural Networks; OK: ordinary Kriging; MLP: multi-layer perceptron     

A bottom‐up approach to discover transition rules of cellular automata using ant intelligence

This paper presents a new method to discover transition rules of geographical cellular automata (CA) based on a bottom‐up approach, ant colony optimization (ACO). CA are capable of simulating the evolution of complex geographical phenomena. The core of a CA model is how to define transition rules so that realistic patterns can be simulated using empirical data. Transition rules are often defined by using mathematical equations, which do not provide easily understandable explicit forms. Furthermore, it is very difficult, if not impossible, to specify equation‐based transition rules for reflecting complex geographical processes. This paper presents a method of using ant intelligence to discover explicit transition rules of urban CA to overcome these limitations. This ‘bottom‐up’ ACO approach for achieving complex task through cooperation and interaction of ants is effective for capturing complex relationships between spatial variables and urban dynamics. A discretization technique is proposed to deal with continuous spatial variables for discovering transition rules hidden in large datasets. The ACO–CA model has been used to simulate rural–urban land conversions in Guangzhou, Guangdong, China. Preliminary results suggest that this ACO–CA method can have a better performance than the decision‐tree CA method.     

Improving map generalisation with new pruning heuristics

Many automated generalisation methods are based on local search optimisation techniques: Starting from an initial state of the data, one or several new child states are produced using some transformation algorithms. These child states are then evaluated according to the final data requirements, and possibly used as new candidate state to transform. According to this approach, the generalisation process can be seen as a walk in a tree, each node representing a state of the data, and each link a transformation. In such an approach, the tree exploration heuristic has a great impact on the final result: Depending on which parts of the tree are either explored or pruned, the final result is different, and the process more or less computationally prohibitive. This article investigates the importance of exploration heuristic choice in automated generalisation. Different pruning criteria are proposed and tested on real generalisation cases. Recommendations on how to choose the pruning criterion depending on the need are provided.     

Performance of Multivariate Adaptive Regression Splines (MARS) in predicting runoff in mid-Himalayan micro-watersheds with limited data / Performances de régressions par splines multiples et adaptives (MARS) pour la prévision d'écoulement au sein de micro-bassins versants Himalayens d'altitudes intermédiaires avec peu de données

Abstract

Steep topography and land-use transformations in Himalayan watersheds have a major impact on hydrological characteristics and flow regimes, and greatly affect the perenniality and sustainability of water resources in the region. To identify the appropriate conservation measures in a watershed properly, and, in particular, to augment flow during lean periods, accurate estimation of streamflow is essential. Due to the complexity of rainfall—runoff relationships in hilly watersheds and non-availability of reliable data, process-based models have limited applicability. In this study, data-driven models, based upon the Multiple Adaptive Regression Splines (MARS) technique, were employed to predict streamflow (surface runoff, baseflow and total runoff) in three mid-Himalayan micro-watersheds. In addition, the effect of length of historical records on the performance of MARS models was critically evaluated. Though acceptable MARS models could be developed with a 2-year data set, their performance improved considerably with a 3-year data set. Various indicators of model performance, such as correlation coefficient, average deviation, average absolute deviation and modelling efficiency, showed significant improvement for simulation of surface runoff, baseflow and total flow. To further analyse the versatility and general applicability of the MARS approach, 2-year data sets were used to develop the model and test it on a third-year data set to assess its performance. The models simulated the surface runoff, baseflow and total flow reasonably well and can be reliably applied in ungauged small watersheds under identical agro-climatic settings.     

Fragility functions of blockwork wharves using artificial neural networks

The use of artificial neural networks in the general framework of a performance-based seismic vulnerability evaluation for earth retaining structures is presented. A blockwork wharf-foundation-backfill complex is modeled with advanced nonlinear 2D finite difference software, wherein liquefaction occurrence is explicitly accounted for. A simulation algorithm is adopted to sample geotechnical input parameters according to their statistical distribution, and extensive time histories analyses are then performed for several earthquake intensity levels. In the process, the seismic input is also considered as a random variable. A large dataset of virtual realizations of the behavior of different configurations under recorded ground motions is thus obtained, and an artificial neural network is implemented in order to find the unknown nonlinear relationships between seismic and geotechnical input data versus the expected performance of the facility. After this process, fragility curves are systematically derived by applying Monte Carlo simulation on the obtained correlations. The novel fragility functions herein proposed for blockwork wharves take into account different geometries, liquefaction occurrence and type of failure mechanism. Results confirm that the detrimental effects of liquefaction increase the probability of failure at all damage states. Moreover, it is also demonstrated that increasing the base width/height ratio results in higher failure probabilities for the horizontal sliding than for the tilting towards the sea.     

Application of statistical learning algorithms to ultimate bearing capacity of shallow foundation on cohesionless soil

This study employs two statistical learning algorithms (Support Vector Machine (SVM) and Relevance Vector Machine (RVM)) for the determination of ultimate bearing capacity (q_u) of shallow foundation on cohesionless soil. SVM is firmly based on the theory of statistical learning, uses regression technique by introducing varepsilon‐insensitive loss function. RVM is based on a Bayesian formulation of a linear model with an appropriate prior that results in a sparse representation. It also gives variance of predicted data. The inputs of models are width of footing (B), depth of footing (D), footing geometry (L/B), unit weight of sand (γ) and angle of shearing resistance (?). Equations have been developed for the determination of q_u of shallow foundation on cohesionless soil based on the SVM and RVM models. Sensitivity analysis has also been carried out to determine the effect of each input parameter. This study shows that the developed SVM and RVM are robust models for the prediction of q_u of shallow foundation on cohesionless soil. Copyright © 2010 John Wiley & Sons, Ltd.     

基于A-star算法的矿井事故救援研究

针对井下不同形状巷道几何空间特征,确立以空间点、线、面为基本图元,以巷道中心线作为三维模型构建的基础框架,建立巷道三维模型。采用启发式路径搜索A-star算法,实现了应急救援路线智能快速选择。实践证明,该方法对于矿井事故定位和救援具有一定的实用性。     

多类支持向量机在语音识别中的应用

针对语音模式识别,引入多类（M-ary）支持向量机来进行分类。并详细的介绍了M-ary支持向量机的概念以及语音识别系统中的实现方式。通过仿真实验,与BP神经网络、SVM进行比较,验证了M-ary支持向量机具有识别率高,抗噪能力强等优越性。     

通用走航式海洋测量作业平台的设计与实现

扼要回顾了我国走航式海洋测量作业平台的研制历程,提出了通用走航式海洋测量作业平台研制的总体目标、设计原则、功能需求,介绍了该平台软件硬件研制实现及其主要性能、功能技术指标,以及平台的应用情况,最后对平台的未来发展进行了展望。     

Modeling the Effects of Meteorological Factors on SO2 and PM10 Concentrations with Statistical Approaches

Air quality has been deteriorated seriously in urban areas as a result of increasing anthropogenic activities. Meteorological conditions affect air pollution levels in the urban atmosphere significantly due to their important role in transport and dilution of the pollutants. This paper aims to investigate usability of some promising statistical methods for examining the impacts of metrological factors on SO₂ and PM10 levels. Data were collected from city centre of Kocaeli in winter periods from 2007 to 2010 as pollutant concentrations increase in winters due to expanding combustion facilities. Results of bivariate correlation analysis showed that humidity and rainfall have remarkable negative correlations with the pollutants. Multiple linear regression models and artificial neural network (ANN) models were used to predict next day's PM10 and SO₂ levels. In regression models calculated R² values were 0.89 and 0.75 for PM10 and SO₂, respectively. Among the various architectures, single layer networks provided better performance in ANN applications. Highest R² values were obtained as 0.89 and 0.69 for PM10 and SO₂, respectively, by using appropriate networks.