app.py

import gradio as gr

from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.inspection import permutation_importance

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

import matplotlib
matplotlib.use('agg')

def create_dataset(num_samples, num_informative):
    X, y = make_classification(
        n_samples=num_samples,
        n_features=10,
        n_informative=num_informative,
        n_redundant=0,
        n_repeated=0,
        n_classes=2,
        random_state=0,
        shuffle=False,
    )

    X_train, X_test, y_train, y_test = train_test_split(X, y, stratify=y, random_state=42)
    return X_train, X_test, y_train, y_test 

def plot_mean_decrease(clf, feature_names):
    importances = clf.feature_importances_
    std = np.std([tree.feature_importances_ for tree in clf.estimators_], axis=0)

    forest_importances = pd.Series(importances, index=feature_names)

    fig, ax = plt.subplots()
    forest_importances.plot.bar(yerr=std, ax=ax)
    ax.set_title("Feature importances using MDI")
    ax.set_ylabel("Mean decrease in impurity")
    fig.tight_layout()

    return fig

def plot_feature_perm(clf, feature_names, X_test, y_test):
    result = permutation_importance(
        clf, X_test, y_test, n_repeats=10, random_state=42, n_jobs=2
    )
    forest_importances = pd.Series(result.importances_mean, index=feature_names)
    
    fig, ax = plt.subplots()
    forest_importances.plot.bar(yerr=result.importances_std, ax=ax)
    ax.set_title("Feature importances using permutation on full model")
    ax.set_ylabel("Mean accuracy decrease")
    fig.tight_layout()

    return fig

def train_model(num_samples, num_info):
    
    X_train, X_test, y_train, y_test = create_dataset(num_samples, num_info)

    feature_names = [f"feature {i}" for i in range(X_train.shape[1])]
    forest = RandomForestClassifier(random_state=0)
    forest.fit(X_train, y_train)

    fig = plot_mean_decrease(forest, feature_names)
    fig2 = plot_feature_perm(forest, feature_names, X_test, y_test)
    return fig, fig2



title = "Feature importances with a forest of trees 🌳"
description = """
            This example shows the use of a random forest model in the evaluation of feature importances \
            of features on an artificial classification task. The model is trained with simulated data that \
            are generated using a user-selected number of informative features. \
            

            The plots show the feature impotances calculated with two different methods. In the first method (left) \
            the importances are provided by the model and they are computed as the mean and standard deviation \
            of accumulation of the impurity decrease within each tree. In the second method (right) uses permutation \
            feature importance which is the decrease in a model score when a single feature value is randomly shuffled. \
            
                
            The blue bars are the feature importances of the random forest model, along with their inter-trees variability \
            represented by the error bars.
            """
            
with gr.Blocks() as demo:
    gr.Markdown(f"## {title}")
    gr.Markdown(description)

    # with gr.Column():
    
    num_samples = gr.Slider(minimum=1000, maximum=5000, step=500, value=1000, label="Number of samples")
    num_info = gr.Slider(minimum=2, maximum=10, step=1, value=3, label="Number of informative features")

    with gr.Row():
        plot = gr.Plot()
        plot2 = gr.Plot()
            
    num_samples.change(fn=train_model, inputs=[num_samples, num_info], outputs=[plot, plot2])
    num_info.change(fn=train_model, inputs=[num_samples, num_info], outputs=[plot, plot2])

    
demo.launch(enable_queue=True)