Statistical Modelling and Variable
Selection in Climate Science
Shalabh and Subhra Sankar Dhar
Abstract Several modelling techniques are used in Statistics to obtain different
models. The method of linear regression analysis is explained in this article. Several
steps starting from concepts, calculation, and interpretation, which are involved in the
modelling process are stepwise explained. The role of ridge regression for choosing
important explanatory variable affecting the outcome is discussed and is used in the
development of LASSO (least absolute shrinkage and selection operator) technique.
How to find the linear regression model and the subset of important variables using
LASSO with an open source R statistical software are illustrated.
Keywords Linear regression model · Variable selection · Model fitting · Ridge
regression · LASSO · Prediction
1 Introduction
Climate and climate science have become important areas of research during the
past decade. This area is intrinsically connected to the survival of living being on this
planet. Nobody can disregard the claim that nature is supreme and has its own laws
to govern the earth. However, in spite of this, human beings are the only creation
of nature who initiated the thought process to understand its rules and phenomenon.
Surely, the nature will never appear before the human being to provide explanations
about its regulations. Nevertheless, the human being never lost the spirit for learning
and attempted to understand the laws and phenomenon of nature by understanding
the various causal factors and variables by moving in the opposite direction. This
direction is to first observe the phenomenon in terms of happenings or non-happening
of events and collect quantified observations on the variables responsible for inputs
Shalabh (B) · S. S. Dhar
Department of Mathematics & Statistics, Indian Institute of Technology,
Kanpur 208016, India
e-mail: shalab@iitk.ac.in
S. S. Dhar
e-mail: subhra@iitk.ac.in
© Springer Nature Switzerland AG 2020
N. Roy et al. (eds.), Socio-economic and Eco-biological Dimensions
in Resource use and Conservation, Environmental Science and Engineering,
https://doi.org/10.1007/978-3-030-32463-6_18
351
Selection in Climate Science
Shalabh and Subhra Sankar Dhar
Abstract Several modelling techniques are used in Statistics to obtain different
models. The method of linear regression analysis is explained in this article. Several
steps starting from concepts, calculation, and interpretation, which are involved in the
modelling process are stepwise explained. The role of ridge regression for choosing
important explanatory variable affecting the outcome is discussed and is used in the
development of LASSO (least absolute shrinkage and selection operator) technique.
How to find the linear regression model and the subset of important variables using
LASSO with an open source R statistical software are illustrated.
Keywords Linear regression model · Variable selection · Model fitting · Ridge
regression · LASSO · Prediction
1 Introduction
Climate and climate science have become important areas of research during the
past decade. This area is intrinsically connected to the survival of living being on this
planet. Nobody can disregard the claim that nature is supreme and has its own laws
to govern the earth. However, in spite of this, human beings are the only creation
of nature who initiated the thought process to understand its rules and phenomenon.
Surely, the nature will never appear before the human being to provide explanations
about its regulations. Nevertheless, the human being never lost the spirit for learning
and attempted to understand the laws and phenomenon of nature by understanding
the various causal factors and variables by moving in the opposite direction. This
direction is to first observe the phenomenon in terms of happenings or non-happening
of events and collect quantified observations on the variables responsible for inputs
Shalabh (B) · S. S. Dhar
Department of Mathematics & Statistics, Indian Institute of Technology,
Kanpur 208016, India
e-mail: shalab@iitk.ac.in
S. S. Dhar
e-mail: subhra@iitk.ac.in
© Springer Nature Switzerland AG 2020
N. Roy et al. (eds.), Socio-economic and Eco-biological Dimensions
in Resource use and Conservation, Environmental Science and Engineering,
https://doi.org/10.1007/978-3-030-32463-6_18
351