Parallel grid search

Parallel Grid Search

Parallelizing grid searches can significantly reduce computation time, especially for independent trials.

Using joblib

joblib is a popular library for parallelizing Python loops.

from joblib import Parallel, delayed
import itertools

def run_experiment(params):
    min_samples, epsilon, shape_weight = params
    # ... logic to run DBSCAN and evaluate ...
    return results

# Define search space
min_samples_range = range(3, 10)
epsilon_range = range(4, 26, 2)
shape_weight_range = [round(x * 0.1, 1) for x in range(9, 20)]

param_combinations = list(itertools.product(min_samples_range, epsilon_range, shape_weight_range))

# Run in parallel
results = Parallel(n_jobs=-1)(delayed(run_experiment)(p) for p in param_combinations)

Considerations

• Data Sharing: Pass only necessary data to workers to minimize overhead.
• Progress Tracking: Use tqdm if a progress bar is needed.
• Resource Management: Be mindful of memory usage when running many parallel processes.