regression3.py

regression3.py

X, y = make_regression(n_samples=100, n_features=1, noise=10, random_state=42)

# Apply a more skewed transformation to the target values (e.g., square root)
y = np.sqrt(np.abs(y)) * X[:,0]**2

%matplotlib qt
import numpy as np
import pandas as pd
import seaborn as sns
from sklearn.datasets import make_regression
from sklearn.dummy import DummyRegressor

# Create a synthetic dataset for regression

CONSTANT = 1
STRATEGIES = ['mean', 'constant']

def plot_dummy_strategy_regressor_1d(
        X, y, target_name="target",
    ):
    dfs = [] # list of df
    # df for input and ground true
    df_true = pd.DataFrame({target_name: y.flatten()})
    df_true['strategy'] = "ground_true"
    df_true['x'] = X[:]
    dfs.append(df_true)

    # add df for each strategy    
    for i, strategy in enumerate(STRATEGIES):
        dummy = DummyRegressor(strategy=strategy, constant=CONSTANT, quantile=QUANTILE)  # Constant value for the 'constant' strategy
        dummy.fit(X, y)
        preds = dummy.predict(X)
        
        df = pd.DataFrame({target_name: preds.flatten()})
        if strategy=="constant":
            strategy=f"constant(={CONSTANT})"
        elif strategy=='quantile':
            strategy=f"quantile(={QUANTILE})"
        strategy+= f"($R^2$={dummy.score(X_test, y_test):.4f})"
        df['strategy'] = strategy
        df['x']=X[:]
        dfs.append(df)
    df = pd.concat(dfs)    
    fg = sns.relplot(data=df, kind='scatter', x='x', y=target_name, hue='strategy')
    return fg, df

fg_1d, df_regressor_1d = plot_dummy_strategy_regressor_1d(X, y, target_name="target")