git/automl_datasets
1
0
Fork 0
Code Issues 1 Pull Requests Actions Packages Projects Releases Wiki Activity

Multi Processing?

Browse Source
This commit is contained in:
Varyngoth
2025-11-11 22:26:10 -04:00
parent c16c545bc8
commit fd0cb67812
1 changed files with 12 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
src/nsga_exp.py
View File

@@ -13,6 +13,7 @@ import shap
from deap import base, creator, tools, algorithms from deap import base, creator, tools, algorithms
from algorithms import lasso, random_forest, gradient_boosting, decision_tree_regressor, ridge_regressor, stacking_lasso from algorithms import lasso, random_forest, gradient_boosting, decision_tree_regressor, ridge_regressor, stacking_lasso
from multiprocessing import Pool, cpu_count
import argparse import argparse
creator.create("FitnessMin", base.Fitness, weights=(-1.0, -1.0)) # Minimize both objectives creator.create("FitnessMin", base.Fitness, weights=(-1.0, -1.0)) # Minimize both objectives
@@ -162,28 +163,33 @@ def run_nsga_experiment(output_base_directory="./results_nsga",
start_time = time.time() start_time = time.time()
pop = toolbox.population(n=population_size) pop = toolbox.population(n=population_size)
# Evaluate the initial population # Use a Pool to parallelize evaluations
fits = toolbox.map(toolbox.evaluate, pop) with Pool(processes=cpu_count()) as pool:
# Evaluate the initial population in parallel
fits = list(pool.map(toolbox.evaluate, pop))
for fit, ind in zip(fits, pop): for fit, ind in zip(fits, pop):
ind.fitness.values = fit[:2] ind.fitness.values = fit[:2]
ind.shap_values = fit[2] ind.shap_values = fit[2]
# Use the number of parents and offspring in the evolution process # Evolution loop
for gen in range(n_generations): for gen in range(n_generations):
if time.time() - start_time > time_limit: if time.time() - start_time > time_limit:
print("Time limit exceeded, stopping evolution.") print("Time limit exceeded, stopping evolution.")
break break
offspring = algorithms.varAnd(pop, toolbox, cxpb=0.7, mutpb=0.2) offspring = algorithms.varAnd(pop, toolbox, cxpb=0.7, mutpb=0.2)
fits = toolbox.map(toolbox.evaluate, offspring)
# Evaluate offspring in parallel
fits = list(pool.map(toolbox.evaluate, offspring))
for fit, ind in zip(fits, offspring): for fit, ind in zip(fits, offspring):
ind.fitness.values = fit[:2] # Ensure only the first two values are assigned to fitness ind.fitness.values = fit[:2]
ind.shap_values = fit[2] ind.shap_values = fit[2]
# Select the next generation of parents from the combined pool of parents and offspring # Select next generation
pop = toolbox.select(pop + offspring, k=population_size) pop = toolbox.select(pop + offspring, k=population_size)
for ind in pop: for ind in pop:
for fold_idx, shap_values in enumerate(ind.shap_values): for fold_idx, shap_values in enumerate(ind.shap_values):
shap_output_path = os.path.join(output_directory, f"shap_values_{int(ind[0])}_fold_{fold_idx + 1}.npy") shap_output_path = os.path.join(output_directory, f"shap_values_{int(ind[0])}_fold_{fold_idx + 1}.npy")