"mode": ['regression'],
"dropout": [0.0, 0.2, 0.4]
}
def gc_model_builder(model_params, model_dir):
gc_model = GraphConvModel(**model_params, model_dir="./models")
return gc_model
i = 0
for train, test in ind:
train_set = dataset.iloc[train]
test_set = dataset.iloc[test]
train_set.to_csv('train_' + str(i) + '.csv')
test_set.to_csv('test_' + str(i) + '.csv')
optimizer = wf.HyperparamOpt(gc_model_builder)
best_model, best_hyperparams, all_results = optimizer.CVgridsearch(
params_dict, train_set)
file = open('opt_result_' + str(i) + '.txt', 'w')
s = 'Best Hyperparameter:' + str(best_hyperparams) + '\n\nAll Results:\n'
file.write(s)
from operator import itemgetter
for k, v in sorted(all_results.items(), key=itemgetter(1)):
s = k + " : " + str(v) + "\n"
file.write(s)
file.close()
if i == 0:
break
i += 1
"learning_rate":[0.005,0.0005, 0.001],
"mode":['regression']
}
def mpnn_model_builder(model_params , model_dir):
return MPNNModel(**model_params, model_dir = "./models")
def gc_model_builder(model_params , model_dir):
gc_model = GraphConvModel(**model_params, model_dir = "./models")
return gc_model
i = 0
for train,test in ind:
if i > 2 && i < 5:
train_set = data.iloc[train]
test_set = data.iloc[test]
train_set.to_csv('train_'+str(i)+'.csv')
test_set.to_csv('test_'+str(i)+'.csv')
optimizer = wf.HyperparamOpt(mpnn_model_builder)
best_model, best_hyperparams, all_results = optimizer.CVgridsearch(mpnn_dict,train_set)
file = open('opt_result_'+str(i)+'.txt', 'w')
s = 'Best Hyperparameter:' + str(best_hyperparams) + '\n\nAll Results:\n'
file.write(s)
from operator import itemgetter
for k, v in sorted(all_results.items(), key=itemgetter(1)):
s = k + " : " + str(v)+"\n"
file.write(s)
file.close()
#if i == 2:
# break
i += 1
