Partial Least Squares Regression Based Variables Selection for Water Level Predictions

Abstract

Floods are common phenomenon in the state of Kuala Krai, specifically in Kelantan-Malaysia. Every year, floods affecting biodiversity on this region and also causing property loss of this residential area. The residents in Kelantan always suffered from floods since the water overflows to the areas adjoining to the rivers, lakes or dams. Months, average monthly rainfall, temperature, relative humidity and surface wind were used as predictors while the water level of Galas River was used as response. The selection of suitable predictor variables becomes an important issue for developing prediction model since the analysis data uses many variables from meteorological and hydrogical departments. In this study, we conduct K-fold Cross-Validation (CV) to select the important variables for the water level predictions. A suitable prediction model is needed to forecast the water level in Galas River by adopting the Ordinary Linear Regression (OLR) and Partial Least Squares Regression (PLSR). However, we need to perform pre-processing data of the datasets since the original data contain missing data. We perform two types of pre-processing data which are using mean of the corresponding months (type I pre-processing data) and OLR (type II pre-processing data) of missing data. Based on the experiment, PLSR is more suitable model rather than OLR for predicting the water level in Galas River and the use of the type I preprocessing data gives higher accuracy than the type II pre-processing data.

Quantifying and Predicting the Water Quality Associated with Land Cover Change: A Case Study of the Blesbok Spruit Catchment, South Africa

“…The PLS analysis is a proven effective variance-based structural equation model approach which is recommended for small sample sizes or in the case when the number of explanatory variables exceeds the number of observation and high level of multicollinearity [44,45]. The use of PLS is recommended as it addresses multi-variable problems such as the ill-effect of multiplex between variables in the case of predicting water quality [45][46][47][48].…”

Predicting water quality associated with land cover change in the Grootdraai Dam catchment, South Africa

“…The PLS analysis is a proven effective variance-based structural equation model approach recommended for small sample sizes or in the case when the number of explanatory variables exceeds the number of observations and a high level of multicollinearity (Ibrahim & Wibowo, 2013a;Nasser & Wisenbaker, 2003). The use of PLS is recommended as it addresses multivariable problems such as the ill-effect of multiplex between variables (Ibrahim & Wibowo, 2013a, 2013bLou, Zhao, Chen, & Zhao, 2009;Singh, Jakubowski, Chidister, & Townsend, 2013).…”

Modelado de las Temperaturas del Aire a 850 milibares: un Potencial Indicador de las Ondas Cálidas en el Noroeste de México

“…En su forma más simple un modelo lineal especifica la relación lineal entre la variable dependiente Y, y un grupo de variables predictivas X (Gonzalez y Alciaturi, 2012). El PLS se ha utilizado para predicción, p.e., Ibrahim et al (2013), lo utilizó para pronosticar el nivel de agua en el río Galas mediante la adopción de la regresión lineal ordinaria y PLS.…”