In this article we discuss the break point existence in the monthly precipitation series collected between January 1965 and December 2005 at two meteorological stations situated on the Romanian Black Sea Littoral. Performing the segmentation procedure of\nHubert, the mDP algorithm and the BP procedure, August and October 2005 have been determined as break points for Mangalia series. Since there is no enough data after October 2005, the model for the precipitation evolution has been designed for the period\nbefore August 2005. The segmentation procedure of Hubert, the mDP algorithm and the BP procedure, on the one hand, and the Pettitt test, one the other hand, provided different change points for Sulina series. Therefore, a Box-Cox transformation has been performed to obtain the data normality, which is a requisite for the Buishand, Lee &\nHeghinian and Barry & Hartighan tests. Since for the transformed series, all the tests gave the same change point (August 1982), alternative models have been built for it, for the periods until and after it. Both approaches (parametrical and nonparametric)\nproposed by us, suggest the same trend for the precipitation evolution at the studied stations.

It is an easily noticed fact that a new generation of residents has been establishing new habitation structures all over Romania. This also applies to the south-eastern Oradea Suburban Area, in north-western Romania. The analysed suburban landscapes bring out open attitudes in former city dwellers, strongly biased pro-landscape (78.52% of all interviewees). The landscape criterion ranks second in reasons for relocation, indicating that local nature meets the expectations of the new residents. Indubitable spiritual benefits are also involved, the new residents’ perception of local landscapes being dominated by responses like beauty, repose, naturalness. However, the new residents do not have a narrowed-down, specialised definition in their minds when expressing opinions on local landscape physiognomy in detail, and on outstanding features that render local landscapes attractive. Even the landscape management interventions of new residents and of local authorities revolve around land estate categories and tailored urbanistic requisitions. Consequently, the configuration of neo-landscapes with a distinct suburban identity emerges. The major directions of this case study may serve as groundwork for further studies on the issue of landscape as subject matter in attracting city dwellers to suburban locations.

Cancer classification has become an active area of research in the biomedical domain. With the increase in the number of cancer victims, an efficient technique for cancer classification is necessary to reduce the death rate of the cancer patients. Several researches are being done to improve the accuracy of cancer classification and also to reduce the convergence time of the algorithms. Cancer classification results in better diagnosis which reduces the death rate of the cancer patients. Moreover, cancer classification plays a vital role in the discovery of drug in the field of medical sciences. In this research, efficient neural network techniques and statistical ranking techniques are used along with able learning algorithms for providing significant cancer classification. The primary objective is to propose efficient cancer classification techniques which provide reliable and significant classification accuracy. The proposed cancer classification approaches have two steps. In the first step, all genes in the training dataset are ranked using a scoring scheme. From this step, the genes with the highest ranks are identified. In the second step, the classification capability of all simple two gene combinations among the genes selected is tested using a neural network algorithm. In this paper uses three proficient classifiers such as Back Propagation Algorithm, Fuzzy Neural Networks and Adaptive Neuro Fuzzy Inference Systems. The performances of the proposed approaches are evaluated on lymphoma datasets. The performance is evaluated based on the performance measures such as classification accuracy, classification time and convergence behavior. From the experimental results, it is observed that the proposed Fast Adaptive Neuro Fuzzy Inference Systems (FANFIS) approach provides significant performance with very high accuracy, less classification time and less convergence time compared to other methods.

The main objective of this paper is at developing an effective machine learning techniques for cancer classification, which could provide reliable cancer classification with better accuracy. The work comprises of two steps. In the first step, important genes are chosen using ANalysis Of VAriance (ANOVA) ranking scheme. The second step involves the classification task using an efficient classifier. Here we use the three efficient classifiers such as Fast Support Vector Machine Learning (FSVML), Fast Extreme Learning Machine Learning (FELML) and Relevance Vector Machine Learning (RVMMLML). The experimental values are computed using three datasets namely Lymphoma, Leukemia and SRBCT. The results are interpreted in terms of Testing Accuracy and Training Time. From the experimental results, it is observed that the proposed RVMMLML approach provides better classification accuracy results compared to other classification on the datasets considered.