features = []
f = open(data_features_file, 'r+')
for line in f:
features.append(line.strip())
f.close()
return features
def get_data():
data = np.loadtxt(fname=data_training_file, delimiter=',')
return data
if __name__ == '__main__':
comm = MPI.COMM_WORLD
rank = comm.Get_rank()
if rank == p_root:
data = get_data()
features = get_features()
else:
data, features = None, None
trees = parallel_create_random_forest(comm, rank, data, features)
if rank == p_root:
forest = RandomForest(trees)
forest.dump(data_classifier_file)
def get_features():
features = []
f = open(data_features_file, 'r+')
for line in f:
features.append(line.strip())
f.close()
return features
if __name__ == '__main__':
forest = RandomForest.load(forest_file)
total_diff = 0
errors = 0
reviews = get_reviews()
off_by = [0] * 5
for review in reviews:
answer = forest.classify(review)
if answer != review['star']:
diff = abs(answer - float(review['star']))
off_by[int(diff)] += 1
#print "Answer: %f, Star: %f, Diff: %f" %(answer, float(review['star']), diff)
errors += 1
total_diff += diff
def get_features(): features = [] f = open(data_features_file, 'r+') for line in f: features.append(line.strip()) f.close() return features if __name__ == '__main__': forest = RandomForest.load(forest_file) total_diff = 0 errors = 0 reviews = get_reviews() off_by = [0]*5 for review in reviews: answer = forest.classify(review) if answer != review['star']: diff = abs(answer-float(review['star'])) off_by[int(diff)] += 1 #print "Answer: %f, Star: %f, Diff: %f" %(answer, float(review['star']), diff) errors += 1 total_diff += diff
def get_features(): features = [] f = open(data_features_file, 'r+') for line in f: features.append(line.strip()) f.close() return features def get_data(): data = np.loadtxt(fname=data_training_file, delimiter=',') return data if __name__ == '__main__': comm = MPI.COMM_WORLD rank = comm.Get_rank() if rank == p_root: data = get_data() features = get_features() else: data, features = None, None trees = parallel_create_random_forest(comm, rank, data, features) if rank == p_root: forest = RandomForest(trees) forest.dump(data_classifier_file)
