R package programming which allows to perform a PLS regression on a dataset
Deliverables:
-
Design of a Partial Least Squares Discriminant Analysis algorithm (PLSDA) with some native S3 functions (print() and summary()), overloaded to display the main results in one-time shot
-
Descriptive and Explanatory Graphs with RShiny Dashboard
-
A detailed tutorial both to understand the plsda and how to manipulate plsda.fit function and plsda.predict function (cf.Wiki in the middle of Github-project navigation bar, at the top of the screen
) -
A detailed tutorial to know how to manipulate Rshiny App cf.
-
A report explaining the PLS regression from a mathematical point of view, with the naming of bibliographical sources.
**Partial Least Squares(PLS) in a few bullet-points **
-
a statistical method that bears some relation to principal components regression; instead of finding hyperplanes of minimum variance between the response and independent variables, it finds a linear regression model by projecting the predicted variables and the observable variables to a new space.
-
Because both the X and Y data are projected to new spaces, the PLS family of methods are known as bilinear factor models.
-
Partial least squares Discriminant Analysis (PLS-DA) is a variant used when the Y is categorical.
-
PLS regression, like PCA, seeks to find components which maximize the variability of predictors but differs from PCA as PLS requires the components to have maximum correlation with the response. PLS is a supervised procedure whereas PCA is unsupervised.
-
In order to explain Y variables from X explanatory variables, the descriptors are summarized into a series of 2X2 orthogonal factors th (factorial axes, latent variables, X-scores). As well, Y variables / targets to be explained and predicted are summarized into a series of components uh (scores Y). During the modeling process, the PLS regression will construct the series of factors (uh, th) so that their covariance is maximum. The number of factors cannot exceed the number of explanatory variables.
You can directly import our package for R from GitHub with Devtools :: , or by downloading the attached package.zip
library(devtools)
install_github("'https://github.com/HugoUrba/PLS-DA-projet-R.git")
library(PLS)
- No dataset to deal with ? Do not worry. The dataset used for our tutorial is attached.