def dtc_predict_actual(data): # split the data into training and testing training_set, testing_set = splitdata_train_test(data, 0.7) # generate the feature and targets for the training and test sets features_training, targets_training = generate_features_targets(training_set) features_testing, targets_testing = generate_features_targets(testing_set) # instantiate a decision tree classifier dtc = DecisionTreeClassifier() # train the classifier dtc.fit(features_training, targets_training) # get predictions predictions = dtc.predict(features_testing) # return the predictions and targets return predictions, targets_testing
def rf_predict_actual(data, n_estimators):
# generate the features and targets
features, targets = generate_features_targets(data)
# instantiate a random forest classifier
rfc = RandomForestClassifier(n_estimators=n_estimators)
# get predictions using 10-fold cross validation with cross_val_predict
predicted = cross_val_predict(rfc, features, targets, cv=10)
# return the predictions and their actual classes
return predicted, targets
if __name__ == "__main__":
data = np.load('galaxy_catalogue.npy')
features, targets = generate_features_targets(data)
# Print the shape of each array to check the arrays are the correct dimensions.
print("Features shape:", features.shape)
print("Targets shape:", targets.shape)
# fraction of data which should be in the training set
fraction_training = 0.7
# split the data using your function
training, testing = splitdata_train_test(data, fraction_training)
# print the key values
print('Number data galaxies:', len(data))
print('Train fraction:', fraction_training)
print('Number of galaxies in training set:', len(training))
print('Number of galaxies in testing set:', len(testing))
predicted_class, actual_class = dtc_predict_actual(data)
# Print some of the initial results
print("Some initial results...\n predicted, actual")
for i in range(10):
print("{}. {}, {}".format(i, predicted_class[i], actual_class[i]))
# get the predicted and actual classes
number_estimators = 50 # Number of trees
predicted, actual = rf_predict_actual(data, number_estimators)
# calculate the model score using your function
accuracy = calculate_accuracy(predicted, actual)
print("Accuracy score:", accuracy)
def rf_predict_actual(data, n_estimators):
features, targets = generate_features_targets(data)
rfc = RandomForestClassifier(n_estimators=n_estimators)
predict = cross_val_predict(rfc, features, targets, cv=10)
return predict, data['class']
def dtc_predict_actual(data):
# split the data into training and testing sets using a training fraction of 0.7
training, testing = splitdata_train_test(data, 0.7)
# generate the feature and targets for the training and test sets
# i.e. train_features, train_targets, test_features, test_targets
train_features, train_targets = generate_features_targets(training)
test_features, test_targets = generate_features_targets(testing)
# instantiate decision tree classifier
dtc = DecisionTreeClassifier()
# train the classifier with the train_features and train_targets
dtc.fit(train_features, train_targets)
# get predictions for the test_features
predictions = dtc.predict(test_features)
# return the predictions and the test_targets
return(predictions, test_targets)
def rf_predict_actual(data, n_estimators):
# generate the features and targets
features, targets = generate_features_targets(data)
# instantiate a random forest classifier using n estimators
rfc = RandomForestClassifier(n_estimators=n_estimators)
# get predictions using 10-fold cross validation with cross_val_predict
predicted = cross_val_predict(rfc, features, targets, cv=10)
# return the predictions and their actual classes
return predicted, targets
from sklearn.model_selection import cross_val_predict
from sklearn.tree import DecisionTreeClassifier
from support_functions import plot_confusion_matrix, generate_features_targets
# Implement the following function
def calculate_accuracy(predicted, actual):
correct = predicted[predicted == actual]
return len(correct)/len(predicted)
if __name__ == "__main__":
data = np.load('galaxy_catalogue.npy')
# split the data
features, targets = generate_features_targets(data)
# train the model to get predicted and actual classes
dtc = DecisionTreeClassifier()
predicted = cross_val_predict(dtc, features, targets, cv=10)
# calculate the model score using your function
model_score = calculate_accuracy(predicted, targets)
print("Our accuracy score:", model_score)
# calculate the models confusion matrix using sklearns confusion_matrix function
class_labels = list(set(targets))
model_cm = confusion_matrix(y_true=targets, y_pred=predicted, labels=class_labels)
# Plot the confusion matrix using the provided functions.
