Compare multiple algorithms

roc_curve.py

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib
import matplotlib.pyplot as plt
from sklearn.metrics import accuracy_score, classification_report, RocCurveDisplay
from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import PCA
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.neural_network import MLPClassifier
from sklearn.neighbors import KNeighborsClassifier
from sklearn.model_selection import train_test_split

# class_row = 'Outcome'
# filename = './diabetes.csv'

class_row = 'DEATH_EVENT'
filename = './heart_failure_clinical_records_dataset.csv'

data = pd.read_csv(filename)

filtered = data.drop(class_row, axis=1)
scaler = StandardScaler()
b_standard = scaler.fit_transform(filtered)

rocs = []

algos = (
    ('K-NN', KNeighborsClassifier, {}),
    ('Decision Tree', DecisionTreeClassifier, {}),
    ('Random Forest', RandomForestClassifier, {}),
    ('MLP', MLPClassifier, {'max_iter':500, 'activation':'logistic'}),
)

print(b_standard.shape)

plt.subplots(2, 2)
plt.figure(figsize=(10.0, 10.0))
for i, (name, klass, args) in enumerate(algos):
    ax = plt.subplot(2, 2, i + 1)
    plt.title(name)

    for i in range(1, min(10, b_standard.shape[1])):
        print(f'Processing: Method {name}, comps {i}')

        pca_model = PCA(n_components=i)
        pca_data_standard = pca_model.fit_transform(b_standard)

        pca_dataset = pd.DataFrame(data=pca_data_standard, columns=[f'PC{j + 1}' for j in range(i)])
        pca_dataset[class_row] = data[class_row]

        x = pca_dataset.drop(class_row, axis=1)
        y = pca_dataset[class_row]
        train_x, test_x, train_y, test_y = train_test_split(x, y, test_size=0.2)

        classifier = klass(**args)
        classifier.fit(train_x, train_y)
        pred = classifier.predict(test_x)

        roc_disp = RocCurveDisplay.from_estimator(
                classifier, 
                test_x, 
                test_y, 
                name=f"{i} components",
                ax=ax,
                alpha=0.75,
        )

        ax.legend(fontsize='xx-small')
        rocs.append(roc_disp)
    
plt.savefig(f'{filename[:-4]}-roc.png')