China
Africa
Explore chapters and articles related to this topic
Dimensionality Reduction and Feature Selection
Published in Sing-Tze Bow, Pattern Recognition and Image Preprocessing, 2002
The objective of the canonical analysis (or principal component analysis) is to derive a linear transformation that will emphasize the difference among the pattern samples belonging to different categories. In other words, the principal component analysis is to define new coordinate axes in directions of high information content useful for classification purposes.
Study and analysis of the streamflow decline in North Algeria
Published in Journal of Applied Water Engineering and Research, 2021
Salima Charifi Bellabas, Saadia Benmamar, Abdellatif Dehni
The application of the MK and Pettitt statistical tests, for the observed annual, seasonal and monthly rainfall, reveals no substantial changes in the rainfall, over the studied period and for all observed series, noted by the total absence of trends and breakpoints in the precipitation series. These research results carried out over the period 1968–2013, are in agreement with the researches realized by Chaouche et al. (2010), Zeroual et al. (2016), Taibi et al. (2015) and Zerouali et al. (2015). According to Zeroual et al. 2016, no significant long-term trend was observed in the Algiers watershed by applying the MK test over the study period 1970–2013. This is probably related and influenced by the large-scale atmospheric index, such as the Southern Oscillation Index (SOI). Which shows an excellent correlation with precipitation in North Algeria, by CAC Canonical Analysis of Correlations (Zeroual et al. 2016).
Multidimensional joint coupling: a case study visualisation approach to movement coordination and variability
Published in Sports Biomechanics, 2020
Gareth Irwin, David G. Kerwin, Genevieve Williams, Richard E. A. Van Emmerik, Karl M. Newell, Joseph Hamill
Movement science has become increasingly interested in the problem of coordination, control and skill. As a result, there has been an increased use of multivariate statistics, non-linear dynamics and network analyses to the problems of system decomposition. For example, the linear statistics of principal component analysis (Daffertshofer, Lamoth, Meijer, & Beek, 2004; Lamoth, Daffertshofer, Huys, & Beek, 2009), canonical analysis (Ivanovic & Ivanovic, 2011; Kakebeeke et al., 2014) and cluster analysis (Sailer, Engert, Dietrich, & Straube, 2000) have been used to decompose the multivariate relations in movement and posture tasks. Stergiou (2004) has introduced some analytic tools from non-linear methods to the analysis of human movement. There have also been non-linear network machine learning approaches to motor control through support vector machines (Chow, Davids, Button, & Rein, 2008).
Ranking the importance of benthic metrics and environmental stressors from over a decade of bioassessment multiple stressor studies in five California waterbodies
Published in Journal of Environmental Science and Health, Part A, 2019
Lenwood W. Hall, Raymond W. Alden, Ronald D. Anderson, William D. Killen
For canonical analysis, the correlation between benthic metrics or environmental variables and their respective canonical variates (CVs) were used as weighting factors in the ranking process. This was done in a manner that is analogous to the weighting done for the squared loadings in PCA. The squared correlation is the proportion of an each variable that is explained by a CV. Only variables with squared correlations >0.10 were considered. In practice, the squared loadings (for PCs) or squared correlations (for CVs) were multiplied by the respective R2 values for ranking. In the case of the canonical correlations analysis (CCA) with PCs, the R2 values were multiplied by both the squared correlations and the squared PC loadings. The products of these calculations were ranked from highest to lowest, with the highest ones given the lowest (top) ranks. Consensus rankings for any given statistical approach were determined by averaging all of the rankings for each benthic metric or BioPC or each environmental variable or EnvPC and creating a ranking system from the lowest to highest of the average rankings.