# -*- coding: utf-8 -*-
from pandas import concat
from plotnine import ggplot, aes, geom_point, geom_text, geom_segment, scale_color_manual, arrow
#interns functions
from .utils import check_is_valid_axis, check_is_valid_geom
from .fviz import overlap_coord, fviz_circle, set_axis, list_colors
from .fviz_dist import fviz_dist
[docs]
def fviz_plsr_ind(
obj,
axis = [0,1],
geom_ind = ("point","text"),
repel = False,
point_args_ind = dict(shape = "o", size = 1.5),
text_args_ind = dict(size = 8),
add_group = True,
geom_group = ("point","text"),
point_args_group = dict(shape = "^", size = 3),
text_args_group = dict(size = 11.5),
palette = None,
x_lim = None,
y_lim = None,
x_label = None,
y_label = None,
title = None,
add_hline = True,
add_vline = True,
add_grid = True,
ggtheme = None
):
"""
Visualize Partial Least Squares Regression - Graph of individuals
:class:`~discrimintools.fviz_plsr_ind` provides plotnine based elegant visualization of :class:`~discrimintools.CPLS`, :class:`~discrimintools.PLSDA`, :class:`~discrimintools.PLSLDA` and :class:`~discrimintools.PLSLOGIT` outputs for individuals.
Parameters
----------
obj : class
An instance of class :class:`~discrimintools.CPLS`, :class:`~discrimintools.PLSDA`, :class:`~discrimintools.PLSLDA`, :class:`~discrimintools.PLSLOGIT`.
axis : list, defaul=[0,1]
Dimensions to be plotted.
geom_ind : str, list or tuple, default = ('point','text')
Geometry to be used for the graph. Possible values are the combinaison of ["point","text"].
- 'point' to show only points,
- 'text' to show only labels,
- ('point','text') to show both types.
repel : bool, default=False
To avoid overplotting text labels.
point_args_ind : dict, default=dict(shape = "o", size = 1.5)
Keywords arguments for `geom_point <https://plotnine.org/reference/geom_point.html>`_.
text_args_ind : dict, default=dict(size = 8)
Keywords arguments for `geom_text <https://plotnine.org/reference/geom_text.html>`_.
add_group : bool, default = True
To show group coordinates.
geom_group : str, list or tuple, default = ('point','text')
See ``geom_ind``.
point_args_group : dict, default = dict(shape = "^", size = 3)
See ``point_args_ind``.
text_args_group : dict, default=dict(size = 11.5)
See ``text_args_ind``.
palette : None or list, default=None
Color palette to be used for coloring by groups.
x_lim : None, list or tuple, default=None
The range of the plotted ``x`` values.
y_lim : None, list or tuple, default=None
The range of the plotted ``y`` values.
x_label : None or str, default = None
The label text of ``x``.
y_label : None or str, default = None
The label text of ``y``.
title : None or str, default = None
The title of the graph you draw.
add_hline : bool, default = True
To add a horizontal line.
add_vline : bool, default = True
To add a vertical line.
add_grid : bool, default=True
To add grid customization.
ggtheme : function, default = None
Plotnine `theme <https://plotnine.org/guide/themes-premade.html>`_ name.
Returns
-------
p : class
A object of class ggplot.
See also
--------
:class:`~discrimintools.fviz_plsr`
Visualize Partial Least Squares Regression (CPLS, PLSDA, PLSLDA, PLSLOGIT).
:class:`~discrimintools.fviz_plsr_var`
Visualize Partial Least Squares Regression (CPLS, PLSDA, PLSLDA, PLSLOGIT) - Graph of variables.
:class:`~discrimintools.fviz_dist`
Visualize distance between barycenter.
Examples
--------
>>> from discrimintools.datasets import load_dataset
>>> from discrimintools import CPLS, fviz_plsr_ind
>>> D = load_dataset("breast") # load traning data
>>> y, X = D["Class"], D.drop(columns=["Class"]) # split into X and y
>>> clf = CPLS()
>>> clf.fit(X,y)
CPLS()
>>> p = fviz_plsr_ind(clf) # graph of individuals
>>> print(p)
.. figure:: ../../../../_static/fviz_plsr_ind.png
Graph of individuals - CPLS
"""
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#check if obj is an instance of class CPLS, PLSDA, PLSLDA, PLSLOGIT
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
if obj.model_ not in ["cpls","plsda","plslda","plslogit"]:
raise TypeError("'obj' must be an instance of class CPLS, PLSDA, PLSLDA, PLSLOGIT")
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#check if valid iaxis
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
check_is_valid_axis(obj,axis)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#check if valid geom
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
check_is_valid_geom(geom_ind,('point','text'))
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#set text arguments - add overlap arguments
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
if repel and "text" in geom_ind:
text_args_ind = dict(**text_args_ind,adjust_text=dict(arrowprops=dict(arrowstyle='-',lw=1.0)))
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#set index and palette
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#set index
index = obj.call_.classes
#set palette
if palette is None:
palette = list_colors[:len(index)]
elif not isinstance(palette,(list,tuple)):
raise TypeError("'palette' must be a list or a tuple of colors")
elif len(palette) != len(index):
raise TypeError(f"'palette' must be a list or tuple with length {len(index)}.")
#set x_text and y_text
x_text, y_text = "Can{}".format(axis[0]+1), "Can{}".format(axis[1]+1)
#concatenate individuals coordinates with target variable
coord = concat((obj.ind_.coord,obj.call_.y),axis=1)
#initialize
p = ggplot(data=coord,mapping=aes(x=x_text,y=y_text,label=coord.index))
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#add individuals coordinates
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
if "point" in geom_ind:
p = p + geom_point(aes(color=obj.call_.target),**point_args_ind)
if "text" in geom_ind:
p = p + geom_text(aes(color=obj.call_.target),**text_args_ind)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#add classes coordinates
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
if add_group:
#check if valid geom
check_is_valid_geom(geom_group,('point','text'))
#set text arguments - add overlap arguments
if repel and "text" in geom_group:
text_args_group = dict(**text_args_group,adjust_text=dict(arrowprops=dict(arrowstyle='-',lw=1.0)))
#classes coordinates
class_coord = obj.classes_.coord
class_coord[f"{obj.call_.target}"] = list(class_coord.index)
#add points
if "point" in geom_group:
p = p + geom_point(data=class_coord,mapping=aes(x=x_text,y=y_text,color=obj.call_.target,label=class_coord.index),**point_args_group)
#add texts
if "text" in geom_group:
p = p + geom_text(data=class_coord,mapping=aes(x=x_text,y=y_text,color=obj.call_.target,label=class_coord.index),**text_args_group)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#add color scale
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
p = p + scale_color_manual(values=palette,labels=index)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#add others elements
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#set x label
if x_label is None:
x_label = "Can{} ({}%)".format(axis[0]+1,round(obj.explained_variance_.iloc[axis[0],0],1))
#set y label
if y_label is None:
y_label = "Can{} ({}%)".format(axis[1]+1,round(obj.explained_variance_.iloc[axis[1],0],1))
#set title
if title is None:
title = "Graph of individuals - {}".format(obj.__class__.__name__)
p = set_axis(p=p,x_lim=x_lim,y_lim=y_lim,x_label=x_label,y_label=y_label,title=title,add_hline=add_hline,add_vline=add_vline,add_grid=add_grid,ggtheme=ggtheme)
return p
[docs]
def fviz_plsr_var(
obj,
axis = [0,1],
geom = ("arrow","text"),
repel = False,
col_var = "steelblue",
segment_args = dict(linetype="solid",size=0.5,alpha=1),
text_args = dict(size=8),
add_circle = True,
col_circle = "gray",
x_lim = (-1.1,1.1),
y_lim = (-1.1,1.1),
x_label = None,
y_label = None,
title = None,
add_hline = True,
add_vline = True,
add_grid = True,
ggtheme = None
):
"""
Visualize Partial Least Squares Regression - Graph of variables
:class:`~discrimintools.fviz_plsr_var` provides plotnine based elegant visualization of :class:`~discrimintools.CPLS`, :class:`~discrimintools.PLSDA`,:class:`~discrimintools.PLSLDA` and :class:`~discrimintools.PLSLOGIT` outputs for variables.
Parameters
----------
obj : class
An instance of class :class:`~discrimintools.CPLS`, :class:`~discrimintools.PLSDA`, :class:`~discrimintools.PLSLDA` or :class:`~discrimintools.PLSLOGIT`.
axis : list, defaul = [0,1]
Dimensions to be plotted.
geom : str, list or tuple, default = ('arrow','text')
Geometry to be used for the graph. Possible values are the combinaison of ["arrow","text"].
- 'arrow' to show only arrows,
- 'text' to show only labels,
- ('arrow','text') to show both types.
repel : bool, default = False
To avoid overplotting text labels.
col_var : str, default = 'steelblue'
Color for the variables segments and texts.
segments_args : dict, default = dict(linetype="solid",size=0.5,alpha=1)
Keywords arguments for `geom_segment <https://plotnine.org/reference/geom_segment.html>`_.
text_args : dict, default = dict(size = 8)
Keywords arguments for `geom_text <https://plotnine.org/reference/geom_text.html>`_.
add_circle : bool, default = True
To draw circle.
col_circle : str, default = "gray"
Color for the circle.
x_lim : None, list or tuple, default = (-1.1,1.1)
The range of the plotted ``x`` values.
y_lim : None, list or tuple, default = (-1.1,1.1)
The range of the plotted ``y`` values.
x_label : None or str, default = None
The label text of ``x``.
y_label : None or str, default = None
The label text of ``y``.
title : None or str, default = None
The title of the graph you draw.
add_hline : bool, default = True
To add a horizontal line.
add_vline : bool, default = True
To add a vertical line.
add_grid : bool, default = True
To add grid customization.
ggtheme : function, default = None
Plotnine `theme <https://plotnine.org/guide/themes-premade.html>`_ name.
Returns
-------
p : class
A object of class ggplot.
See also
--------
:class:`~discrimintools.fviz_plsr`
Visualize Partial Least Squares Regression (CPLS, PLSDA, PLSLDA, PLSLOGIT)
:class:`~discrimintools.fviz_plsr_ind`
Visualize Partial Least Squares Regression (CPLS, PLSDA, PLSLDA, PLSLOGIT) - Graph of individuals.
:class:`~discrimintools.fviz_dist`
Visualize distance between barycenter.
Examples
--------
>>> from discrimintools.datasets import load_dataset
>>> from discrimintools import CPLS, fviz_plsr_var
>>> D = load_dataset("breast") # load traning data
>>> y, X = D["Class"], D.drop(columns=["Class"]) # split into X and y
>>> clf = CPLS()
>>> clf.fit(X,y)
CPLS()
>>> p = fviz_plsr_var(clf) # graph of variables
>>> print(p)
.. figure:: ../../../../_static/fviz_plsr_var.png
Graph of variables - CPLS
"""
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#check if obj is an instance of class CPLS, PLSDA, PLSLDA, PLSLOGIT
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
if obj.model_ not in ["cpls","plsda","plslda","plslogit"]:
raise TypeError("'obj' must be an instance of class CPLS, PLSDA, PLSLDA, PLSLOGIT")
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#check if valid axis
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
check_is_valid_axis(obj,axis)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#check if valid geom
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
check_is_valid_geom(geom,choice=("arrow","text"))
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#define text loadings - overlap texts
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
coord = overlap_coord(coord=obj.var_.loadings,x_name="Can"+str(axis[0]+1),y_name="Can"+str(axis[1]+1),repel=repel)
#initialize
p = ggplot(data=coord,mapping=aes(x=f"Can{axis[0]+1}",y=f"Can{axis[1]+1}",label=coord.index))
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#add variables loadings
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#add arrows
if "arrow" in geom:
p = p + geom_segment(aes(x=0,y=0,xend=f"Can{axis[0]+1}",yend=f"Can{axis[1]+1}"),color=col_var, **segment_args, arrow = arrow(angle=30,length=0.2/2.54,type="open"))
#dd texts
if "text" in geom:
p = p + geom_text(aes(x="xnew",y="ynew"),color=col_var,**text_args)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#add correlation circle
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
if add_circle:
p = fviz_circle(p=p,color=col_circle)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#add others elements
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#set x label
if x_label is None:
x_label = "Can{} ({}%)".format(axis[0]+1,round(obj.explained_variance_.iloc[axis[0],0],1))
#set y label
if y_label is None:
y_label = "Can{} ({}%)".format(axis[1]+1,round(obj.explained_variance_.iloc[axis[1],0],1))
#set title
if title is None:
title = "Graph of variables - {}".format(obj.__class__.__name__)
p = set_axis(p=p,x_lim=x_lim,y_lim=y_lim,x_label=x_label,y_label=y_label,title=title,add_hline=add_hline,add_vline=add_vline,add_grid=add_grid,ggtheme=ggtheme)
return p
[docs]
def fviz_plsr(
obj, element = "ind", **kwargs
):
"""
Visualize Partial Least Squares Regression (CPLS, PLSDA, PLSLDA, PLSLOGIT)
:class:`~discrimintools.fviz_plsr` provides plotnine based elegant visualization of :class:`~discrimintools.CPLS`, :class:`~discrimintools.PLSDA`, :class:`~discrimintools.PLSLDA` and :class:`~discrimintools.PLSLOGIT` outputs.
Parameters
----------
obj : class
An object of class :class:`~discrimintools.CPLS`, :class:`~discrimintools.PLSDA`, :class:`~discrimintools.PLSLDA`, :class:`~discrimintools.PLSLOGIT`.
element : str, default = 'ind'
The element to plot from the output, possible values:
- 'ind' for the individuals graphs
- 'var' for the variables graphs (= Correlation circle)
- 'dist' for the distance graphs
**kwargs :
further arguments passed to or from other methods
Returns
-------
p : class
A object of class ggplot.
See also
--------
:class:`~discrimintools.fviz_plsr_ind`
Visualize Partial Least Squares Regression (CPLS, PLSDA, PLSLDA, PLSLOGIT) - Graph of individuals.
:class:`~discrimintools.fviz_plsr_var`
Visualize Partial Least Squares Regression (CPLS, PLSDA, PLSLDA, PLSLOGIT) - Graph of variables.
:class:`~discrimintools.fviz_dist`
Visualize distance between barycenter.
Examples
--------
>>> from discrimintools.datasets import load_dataset
>>> from discrimintools import CPLS, fviz_plsr
>>> D = load_dataset("breast") # load traning data
>>> y, X = D["Class"], D.drop(columns=["Class"]) # split into X and y
>>> clf = CPLS()
>>> clf.fit(X,y)
CPLS()
Graph of individuals...
>>> p = fviz_plsr(clf, "ind") # graph of individuals
>>> print(p)
.. figure:: ../../../../_static/fviz_plsr_ind.png
Graph of individuals - CPLS
Graph of variables...
>>> p = fviz_cpls(clf, "var") # graph of variables
>>> print(p)
.. figure:: ../../../../_static/fviz_plsr_var.png
Graph of variables - CPLS
Distance between class barycenter.
>>> p = fviz_cpls(clf, "dist") # graph of distance
>>> print(p)
.. figure:: ../../../../_static/fviz_plsr_dist.png
Distance between barycenter - CPLS
"""
if element == "ind":
return fviz_plsr_ind(obj,**kwargs)
elif element == "var":
return fviz_plsr_var(obj,**kwargs)
elif element == "dist":
return fviz_dist(obj,**kwargs)
else:
raise ValueError("'element' should be one of 'ind', 'var','dist'")
