Benchmark ResidualsPlot performance on a large dataset

residuals.py

import time 
import numpy as np 
import matplotlib.pyplot as plt 

from sklearn.datasets import make_regression 
from sklearn.model_selection import train_test_split as tts

from sklearn.linear_model import LinearRegression 
from yellowbrick.regressor import ResidualsPlot


X, y = make_regression(
    n_samples=150000, n_features=30, noise=0.3, bias=0.32, 
    n_informative=22, tail_strength=0.3, effective_rank=2
)  
X_train, X_test, y_train, y_test = tts(X, y, test_size=0.2)

_, ax = plt.subplots() 

start = time.time() 
oz = ResidualsPlot(LinearRegression()) 
oz.fit(X_train, y_train)
print("fit time was {:0.3f} seconds".format(time.time() - start))

start = time.time() 
oz.score(X_test, y_test) 
print("score time was {:0.3f} seconds".format(time.time()-start))

start = time.time()
oz.poof("residuals.pdf")
print("poof time was {:0.3f} seconds".format(time.time()-start))

I got the following output using PDF (much shorter if rendering PNG):

• fit time was 0.459 seconds
• score time was 0.094 seconds
• poof time was 5.538 seconds