neuralnet_regression.py

import numpy as np
from nnetsauce.neuralnet import NeuralNetRegressor
from sklearn.datasets import load_diabetes, fetch_california_housing
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error
from time import time

datasets = ["diabetes", "california_housing"]

for dataset in datasets:
    print(f"\n\n Dataset: {dataset} -----\n")
    if dataset == "diabetes":
        X, y = load_diabetes(return_X_y=True)
    elif dataset == "california_housing":
        X, y = fetch_california_housing(return_X_y=True)
        X, y = X[:1000], y[:1000]
    X_train, X_test, y_train, y_test = train_test_split(X, y, 
                                                        test_size=0.2, 
                                                        random_state=123)

    X_train_1, X_train_2, y_train_1, y_train_2 = train_test_split(X_train, y_train, 
                                                                test_size=0.5, 
                                                                random_state=42)

    model = NeuralNetRegressor(hidden_layer_sizes=(100,), 
                            max_iter=100, 
                            learning_rate=0.01, 
                            random_state=42)
    start_time = time()
    model.fit(X_train_1, y_train_1)
    end_time = time()
    print(f"Time taken to fit model: {end_time - start_time:.2f} seconds")

    y_pred = model.predict(X_test)
    mse = mean_squared_error(y_test, y_pred)
    print(f"Root Mean Squared Error: {np.sqrt(mse):.4f}")

    model2 = NeuralNetRegressor(max_iter=100, 
                            learning_rate=0.01, 
                            random_state=42,
                            weights=model.get_weights())
    start_time = time()
    model2.fit(X_train_2, y_train_2)
    end_time = time()
    print(f"Time taken to fit model: {end_time - start_time:.2f} seconds")

    y_pred = model2.predict(X_test)
    mse = mean_squared_error(y_test, y_pred)
    print(f"Root Mean Squared Error: {np.sqrt(mse):.4f}")