Advances in variable selection methods I: Causal selection methods versus stepwise regression and principal component analysis on data of known and unknown functional relationships

Ssegane, Herbert; Tollner, E. W.; Mohamoud, Yusuf M.; Rasmussen, Todd C.; Dowd, John F.

doi:10.1016/j.jhydrol.2012.01.008

Cited by 53 publications

(27 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Some Evolutionary Computation algorithms were proposed to determine an optimal architecture of ANN (Castillo et al, 2000;Huang et al, 2009) and were applied to hydrological problems ), however they are applicable rather to problems when neither expert-knowledge is available nor physically-based choice of input variables and model complexity is possible. Although a number of other methods how to develop ANN architecture exists (Sietsma and Dow, 1991;Wang et al, 1994;Islam et al, 2009;Ssegane et al, 2012;Nourani and Sayyah Frad, 2012), they usually rely on heuristic or subjective decisions and none is widely applied (Zhang et al, 1998).…”

Section: Introductionmentioning

confidence: 99%

A comparison of methods to avoid overfitting in neural networks training in the case of catchment runoff modelling

Piotrowski

Napiorkowski

2013

Journal of Hydrology

221

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 99%

A comparison of methods to avoid overfitting in neural networks training in the case of catchment runoff modelling

Piotrowski

Napiorkowski

2013

Journal of Hydrology

221

View full text Add to dashboard Cite

“…A two-part study compared the performance of different variable selection 4 methods for clustering basins (Ssegane et al, 2012a) and modeling percentile flows (Ssegane et al, 2012b). These studies 5 revealed the importance of different variables through the variable selection process (i.e.…”

Section: Independent Variables For Regional Regression Modeling 26mentioning

confidence: 99%

Regional regression models of percentile flows for the contiguous US: Expert versus data-driven independent variable selection

Fouad

Skupin

Tague

2016

Preprint

View full text Add to dashboard Cite

Abstract. Percentile flows are statistics derived from the flow duration curve (FDC) that describe the flow equaled or 8 exceeded for a given percent of time. These statistics provide important information for managing rivers, but are often 9 unavailable since most basins are ungauged. A common approach for predicting percentile flows is to deploy regional 10 regression models based on gauged percentile flows and related independent variables derived from physical and climatic 11 data. The first step of this process identifies groups of basins through a cluster analysis of the independent variables, 12 followed by the development of a regression model for each group. This entire process hinges on the independent variables 13 selected to summarize the physical and climatic state of basins. Distributed physical and climatic datasets now exist for the 14 contiguous United States (US). However, it remains unclear how to best represent these data for the development of regional 15 regression models. The study presented here developed regional regression models for the contiguous US, and evaluated the 16 effect of different approaches for selecting the initial set of independent variables on the predictive performance of the 17 regional regression models. An expert assessment of the dominant controls on the FDC was used to identify a small set of 18 independent variables likely related to percentile flows. A data-driven approach was also applied to evaluate two larger sets 19 of variables that consist of either (1) the averages of data for each basin or (2) both the averages and statistical distribution of 20 basin data distributed in space and time. The small set of variables from the expert assessment of the FDC and two larger 21 sets of variables for the data-driven approach were each applied for a regional regression procedure. Differences in 22 predictive performance were evaluated using 184 validation basins withheld from regression model development. The small 23 set of independent variables selected through expert assessment produced similar, if not better, performance than the two 24 larger sets of variables. A parsimonious set of variables only consisted of mean annual precipitation, potential 25 evapotranspiration, and baseflow index. Additional variables in the two larger sets of variables added little to no predictive 26 information. Regional regression models based on the parsimonious set of variables were developed using 734 calibration 27 basins, and were converted into a tool for predicting 13 percentile flows in the contiguous US. Supplementary Material for 28 this paper includes an R graphical user interface for predicting the percentile flows of basins within the range of conditions 29 used to calibrate the regression models. The equations and performance statistics of the models are also supplied in tabular 30 form. 31Hydrol. Earth Syst. Sci. Discuss.,

show abstract

“…PCA uses an orthogonal transformation to convert a set of observations of possibly correlated variables into a set of values of linearly uncorrelated variables. Ssegane et al (2012) compared SRA with PCA and tested both methods' performances on data of known and unknown functional relationships. In nonlinear feature extraction, Guo et al (2011) applied genetic programming (GP) to perform automatic feature extraction from original feature database with the aim of improving the discriminatory performance of a classifier and reducing the input feature dimensionality at the same time.…”

Section: Feature Selection For Dimensionality Reductionmentioning

confidence: 99%

A critical feature extraction by kernel PCA in stock trading model

Chang

2014

Soft Comput

View full text Add to dashboard Cite

This paper presents a kernel-based principal component analysis (kernel PCA) to extract critical features for improving the performance of a stock trading model. The feature extraction method is one of the techniques to solve dimensionality reduction problems (DRP). The kernel PCA is a feature extraction approach which has been applied to data transformation from known variables to capture critical information. The kernel PCA is a kernel-based data mapping tool that has characteristics of both principal component analysis and non-linear mapping. The feature selection method is another DRP technique that selects only a small set of features from known variables, but these features still indicate possible collinearity problems that fail to reflect clear information. However, most feature extraction methods use a variable mapping application to eliminate noisy and collinear variables. In this research, we use the kernel-PCA method in a stock trading model to transform stock technical indices (TI) which allows features of smaller dimension to be formed. The kernel-PCA method has been applied to various stocks and sliding window testing methods using both half-year and 1-year testing strategies. The experimental results show that the proposed method generates more profits than other DRP methods on the America stock market. This stock trading model is very practical for real-world application, and it can be implemented in a real-time environment.

show abstract

Advances in variable selection methods I: Causal selection methods versus stepwise regression and principal component analysis on data of known and unknown functional relationships

Cited by 53 publications

References 34 publications

A comparison of methods to avoid overfitting in neural networks training in the case of catchment runoff modelling

A comparison of methods to avoid overfitting in neural networks training in the case of catchment runoff modelling

Regional regression models of percentile flows for the contiguous US: Expert versus data-driven independent variable selection

A critical feature extraction by kernel PCA in stock trading model

Contact Info

Product

Resources

About