csr_polynomial benchmark

benchmark.py

import numpy as np
from scipy import sparse as sp

def generate_data(n_samples, n_features, n_classes=2, X_density=1, y_sparse=False, dtype=np.float64, random_state=None):
    rng = np.random.RandomState(random_state)
    if X_density < 1:
        X = sp.random(n_samples, n_features, format="csr", density=X_density, random_state=rng)
    else:
        X = np.round(rng.rand(n_samples,n_features)*50).astype(dtype)
    y = np.round(rng.randint(n_classes,size=(n_samples,))).astype(dtype)
    if y_sparse:
        y = sp.csr_matrix(y)
    if y_sparse and y.shape[0] == 1:
        y = y.T
    return X, y

from functools import partial
from time import perf_counter
from statistics import mean, stdev
from itertools import product
import csv
from pathlib import Path
from sklearn.preprocessing import PolynomialFeatures

results_path = 'local_artifacts/benchmarks/csr_polynomial/'
Path(results_path).mkdir(parents=True, exist_ok=True)
branch = "main"


benchmark_config = [
    (
        PolynomialFeatures,
        partial(generate_data, n_samples=2_000),
        product(
            [50, 100, 175],
            [.1, .25, .5],
        ),
    ),
]


N_REPEATS = 30

with open(f'{results_path}{branch}.csv', 'w', newline='') as csvfile:

    writer = csv.DictWriter(
        csvfile,
        fieldnames=[
            "n_features",
            "density",
            "n_repeat",
            "duration",
        ],
    )
    writer.writeheader()

    for Est, make_data, items in benchmark_config:

        for n_features, density in items:
            time_results = []
            for n_repeat in range(N_REPEATS):
                X, _ = make_data(X_density=density, n_features=n_features, random_state=n_repeat)
                est = Est(degree=3)
                start = perf_counter()
                est.fit_transform(X)
                duration = perf_counter() - start
                time_results.append(duration)
                writer.writerow(
                    {
                        "n_features": n_features,
                        "density": density,
                        "n_repeat": n_repeat,
                        "duration": duration,
                    }
                )
            results_mean, results_stdev = mean(time_results), stdev(time_results)
            print(
                f" {n_features=} {density=}|"
                f" {results_mean:.3f} +/- {results_stdev:.3f}"
            )


import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns
plt.rc('font', size=12)
results_path = 'local_artifacts/benchmarks/csr_polynomial/'

_branches = ("main", "PR")
percentile_trim = .95
branches = {br:pd.read_csv(f'{results_path}{br}.csv') for br in _branches}

df = pd.concat([branches[br].assign(branch=br) for br in _branches])

group_by_attrs = ["n_features", "density"]

grouped = list(df.groupby(group_by_attrs))
fig, axis = plt.subplots(3, 3, figsize=(14, 9), constrained_layout=True)
fig.patch.set_facecolor('white')
for (grouped_attrs, subset), ax in zip(grouped, axis.reshape(-1)):

    # Optionally trim outlier data
    if percentile_trim < 1:
        for branch in _branches:
            _subset = subset[subset["branch"]==branch]
            cut = _subset.duration < _subset.duration.quantile(percentile_trim)
            subset[subset["branch"]==branch] = _subset[cut]

    sns.violinplot(data=subset, y="duration", x="branch", ax=ax)
    ax.set_title("|".join( [f"{k}={v}" for k, v in zip(group_by_attrs,grouped_attrs)] ))
    ax.set_xlabel("")
    
for ax in axis[:, 1:].ravel():
    ax.set_ylabel("")