CPLS - breast dataset

#disable warnings
from warnings import simplefilter, filterwarnings
simplefilter(action='ignore', category=FutureWarning)
filterwarnings("ignore")

breast dataset

#alcools dataset
from discrimintools.datasets import load_dataset
D = load_dataset("breast")
D.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 699 entries, 0 to 698
Data columns (total 4 columns):
 #   Column     Non-Null Count  Dtype
---  ------     --------------  -----
 0   ucellsize  699 non-null    int64
 1   normnucl   699 non-null    int64
 2   mitoses    699 non-null    int64
 3   Class      699 non-null    object
dtypes: int64(3), object(1)
memory usage: 22.0+ KB

#split into X and y
y, X = D["Class"], D.drop(columns=["Class"])

Instanciation and training

#instanciation and training
from discrimintools import CPLS
clf = CPLS()
clf.fit(X,y)

CPLS()

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Coefficients

#coefficients
print(clf.coef_.to_frame())

           positive
Constant  -0.424881
ucellsize  0.085323
normnucl   0.053251
mitoses    0.003001

Summary

#summary
from discrimintools import summaryCPLS
summaryCPLS(clf,detailed=True)

                     Partial Least Squares for Classification - Results

Class Level Information:
          Frequency  Proportion  Prior Probability
negative        458      0.6552             0.6552
positive        241      0.3448             0.3448

Importance of PLS components:
      Proportion (%)  Cumulative (%)
Can1         69.1520         69.1520
Can2         20.1981         89.3501

Classification functions coefficients:
           positive     VIP
Constant    -0.4249     NaN
ucellsize    0.0853  1.2038
normnucl     0.0533  1.0346
mitoses      0.0030  0.6933

Classification Summary for Calibration Data:

Observation Profile:
                        Read  Used
Number of Observations   699   699

Number of Observations Classified into Class:
prediction  negative  positive  Total
Class
negative         440        18    458
positive          26       215    241
Total            466       233    699

Percent Classified into Class:
prediction  negative  positive  Total
Class
negative     96.0699    3.9301  100.0
positive     10.7884   89.2116  100.0
Total        66.6667   33.3333  100.0
Priors        0.6552    0.3448    NaN

Error Count Estimates for Class:
        negative  positive   Total
Rate      0.0393    0.1079  0.0629
Priors    0.6552    0.3448     NaN

Classification Report for Class:
              precision  recall  f1-score   support
negative         0.9442  0.9607    0.9524  458.0000
positive         0.9227  0.8921    0.9072  241.0000
accuracy         0.9371  0.9371    0.9371    0.9371
macro avg        0.9335  0.9264    0.9298  699.0000
weighted avg     0.9368  0.9371    0.9368  699.0000

Plotting

#plotting
from discrimintools import fviz_plsr

Graph of individuals

#graph of individuals
p = fviz_plsr(clf,element="ind",repel=False)
p.show()

Graph of variables

#graph of variables
p = fviz_plsr(clf,element="var",repel=True)
p.show()

Distance between barycenter

#distance between barycenter
p = fviz_plsr(clf,element="dist",repel=False,y_lim=(-0.2,0.15))
p.show()