def post(self):
"""Classifies given data using specified classifier."""
# Retrieve specified classifier from database.
classifier_id = self.request.POST['classifier_id']
conn = ds.create_sqlite_connection()
db_entry = ds.fetch_classifier(conn, classifier_id)
classifier = db_entry[3]
# Transform given MRS data for classifier input.
file_name = self.request.POST['myfile'].filename
raw_data = self.request.POST['myfile'].file.read()
d = dataparser.get_xy_data(raw_data)
fftd = fourier_transformer.get_fft(d)
# Classify the transformed MRS data.
test_input = np.array([fftd])
classification = classifier.predict(test_input)
# Show classification results.
template = JINJA_ENVIRONMENT.get_template('classificationresults.html')
self.response.write(template.render(
classification=classification, file_name=file_name))
def prepare_mrs_data_set(self):
"""Retrieves all specified MRS data entries and processes each entry.
Each MRS file is parsed and FFT is applied if specified.
Returns:
Tuple containing (list of sample inputs, list of sample outputs).
"""
training_data_ids = self.request.get_all("training_data_ids")
apply_fft = 'apply_fft' in self.request.POST
LOGGER.debug('Processing MRS data: apply_fft=%s', apply_fft)
# Retrieve specified training data from the database.
conn = ds.create_sqlite_connection()
db_entries = [ds.fetch_mrs_data(conn, data_id) for data_id in training_data_ids]
# Separate each database entry into input and output.
sample_inputs = []
sample_outputs = []
for entry in db_entries:
# Parse data points from the file contents.
mrs_data = dataparser.get_xy_data(str(entry[2]))
# Apply FFT to the data points if specified by user.
if apply_fft:
mrs_data = fourier_transformer.get_fft(mrs_data)
# Add input, output pair to separate lists.
sample_inputs.append(mrs_data)
sample_outputs.append(entry[3])
# Format the data for classifier input.
n_samples = len(sample_inputs)
n_features = len(sample_inputs[0])
sample_inputs = np.array(sample_inputs) # convert before using as buffer
sample_inputs = np.ndarray(
shape=(n_samples, n_features), dtype=float, buffer=sample_inputs)
# Labels for training.
sample_outputs = np.array(sample_outputs)
# Return processed MRS data.
return (sample_inputs, sample_outputs)
def setUpClass(cls):
"""Parse data from a MRS data file."""
mrs_data_string = str(open('data/05_E2', 'r').read())
cls.mrs_data = dataparser.get_xy_data(mrs_data_string)