def tree_selection_heuristic():
# The feature set that is going to be evaluated
feature_set = [0]
max_accuracy = 0.0
best_set = []
best_model = None
# For every set of 1 feature the recursive search is applied
for i in xrange(1, feature_amount):
feature_set.pop()
feature_set.append(i)
new_x_norm, y, _, _ = getDataSubSet(feature_set)
#INIT FOR RECURSIVE CALL
model, calc_best_model, calc_accuracy = manual_cross_validation(
new_x_norm, y, models, names, True)
#RECURSIVE CALL
calc_accuracy, calc_best_set, calc_best_model = recursive_tree_exploration(
feature_set, calc_accuracy, calc_best_model)
if calc_accuracy > max_accuracy:
max_accuracy = calc_accuracy
best_set = calc_best_set
best_model = calc_best_model
print "-----------------------------------"
print "---------BEST FEATURE SET----------"
print best_set
print "-----------------------------------"
print "-------------ACCURACY--------------"
print max_accuracy
return best_model
def recursive_tree_exploration(feature_set, past_accuracy, past_model):
last_added = feature_set[len(feature_set) - 1]
new_feature_set = copy.copy(feature_set)
#POR QUe AGREGAR 0 ??
## COUNT 1 APPEND
new_feature_set.append(0)
max_accuracy = past_accuracy
best_set = feature_set
best_model = past_model
for iter in xrange(last_added + 1, feature_amount):
## COUNT 1 POP
new_feature_set.pop()
## COUNT 1 APPEND
new_feature_set.append(iter)
new_x_norm, y, _, _ = getDataSubSet(new_feature_set)
model, calc_best_model, calc_accuracy = manual_cross_validation(
new_x_norm, y, models, names, True)
if calc_accuracy > max_accuracy:
max_accuracy = calc_accuracy
best_set = new_feature_set
best_model = calc_best_model
calc_accuracy, calc_best_set, calc_best_model = recursive_tree_exploration(
new_feature_set, max_accuracy, calc_best_model)
if calc_accuracy > max_accuracy:
max_accuracy = calc_accuracy
best_set = calc_best_set
best_model = calc_best_model
print best_set
print "BEST SET ACCURACY: " + str(max_accuracy)
print "BEST SET MODEL: " + str(best_model)
return max_accuracy, best_set, best_model
try:
X,y,X_test,Y_test= getDataSubSet(subset)
scaler = preprocessing.MinMaxScaler()
scaler.fit(X)
x_norm=scaler.transform(X)
#initialize models
forest= RandomForestClassifier(n_estimators=100, max_depth=20, random_state=111)
gdBoost = GradientBoostingClassifier(random_state=111)
mlp = MLPClassifier(solver='lbfgs', alpha=1e-5,hidden_layer_sizes=(10, 2), random_state=111)
models=[forest,gdBoost,mlp]
names= ["Random Forest", "Gradient Boosting", "MuliLayer Perceptrons"]
##Using all the features
model,name,mean=manual_cross_validation(x_norm, y, models, names, True)
if(mean>bestMean):
bestModel,bestName,bestMean=model,name,mean
bestSet = subset
except:
print ("Exception raised")
print("###################%s########################"%features_domain[set_number])
print("The best model is: %s with and average accuracy of: %0.5f"%(bestName,bestMean))
print("And with the set of features:")
print (subset)
print("###########")
gdBoost = GradientBoostingClassifier(random_state=111)
mlp = MLPClassifier(solver='lbfgs', alpha=1e-5, hidden_layer_sizes=(10, 2), random_state=111)
models = [forest, gdBoost, mlp]
names = ["Random Forest", "Gradient Boosting", "MuliLayer Perceptrons"]
# Vars to select the best suited model
bestModel = None
bestName = "none"
bestMean = 0.0
x_best = []
scenario = "none"
# Using all the features
print("------------------------------------------")
print("------All Features -----------------------")
model, name, mean = manual_cross_validation(x_norm, y, models, names)
if mean > bestMean:
bestModel, bestName, bestMean = model, name, mean
x_best = x_norm
scenario = "all"
# Removing features with low variance
print("------------------------------------------")
print("------Removing features with low variance -----------------------")
sel = VarianceThreshold(threshold=(0.01))
x_case = sel.fit_transform(x_norm)
model, name, mean = manual_cross_validation(x_case, y, models, names)
if mean > bestMean:
bestModel, bestName, bestMean = model, name, mean
x_best = x_case
scenario = "variance"