def _process_people(self, directory):
files = [f for f in listdir(directory) if isfile(join(directory, f))]
preproc = Preprocessing(DATA_FREQUENCY)
people = []
num_samples = 0
print("Starting preprocessing")
for file in files:
try:
number = self._get_file_number(file)
person = preproc.process_person(
f"{DATA_PATH}/{file}", f"{ORIGINALS_PATH}/s{number}.bdf",
number)
people.append(person)
person_sample_count = len(person)
print(
f"{file} done. Got data from {person_sample_count} videos."
)
num_samples += person_sample_count
print(f"Has {num_samples} samples already")
except Exception:
# raise
pass
print("Preprocessing finished")
return people
def __init__(self, freq, model):
self.freq = freq
self.model = model
self.pre = Preprocessing(self.freq)
self.post = Postprocessing(self.freq)
self.base_bvp_features = None
self.base_gsr_features = None
def _show_statistics(self):
print("Creating statistics...")
main.EXTRACT_ALL_FEATURES = True
x, y = self._get_data_tuples()
labels = Preprocessing.get_labels()
# print(y)
stats = Statistics(x, y, labels)
stats.create()
def resume():
print('start...')
preprocessing = Preprocessing()
vsm = VSM()
svd = SVD()
result = preprocessing.read_file()
judul = preprocessing.read_title()
sentences = preprocessing.split_sentence(result)
sumarize = []
original = []
pre_judul = preprocessing.preprocessing(judul)
index = 0
path = os.path.abspath(os.path.join(os.path.dirname(__file__),
'hasil_uji'))
for s in sentences:
listterm = []
term_fn = []
sentence_result = ''
original_result = ''
pre_result = preprocessing.preprocessing(s)
listterm = vsm.list_term(pre_result)
term_frequency = vsm.cal_tf(pre_result, listterm)
term_frequency_judul = vsm.cal_tf_basic(pre_judul[index], listterm)
term_frequency_normalize = vsm.cal_tf_normalize(term_frequency)
term_frequency_normalize_judul = np.log(
1 + np.asarray(term_frequency_judul))
document_frequency = vsm.cal_df(pre_result, listterm)
document_frequency_judul = vsm.cal_df(pre_judul, listterm)
for idx, tfn in enumerate(term_frequency_normalize):
term_fn.append(tfn.tolist())
weight = vsm.cal_vsm(term_fn, document_frequency)
weight_judul = vsm.cal_vsm_basic(term_frequency_normalize_judul,
document_frequency_judul)
index_sentence = svd.cal_svd(weight, weight_judul)
for i in index_sentence:
sentence_result += s[i] + ' '
for sen in s:
original_result += sen
sumarize.append(sentence_result)
original.append(original_result)
f = open(path + "/berita-" + str(index) + ".txt", "w+")
f.write(sentence_result)
print(index)
index += 1
hasil_ringkasan = []
for i in range(len(sumarize)):
hasil_ringkasan.append({'ringkasan': sumarize[i], 'judul': judul[i]})
try:
mongo.db.ringkasan.insert_many(hasil_ringkasan)
except Exception as e:
print(e)
Exception(e)
return jsonify({"sumarize": sumarize, "original": original}), 200
def _reverse_sbs_scores(self):
print("Looking for best features...")
x, y = self._get_data_tuples()
print(len(y))
ai = AI()
features, accuracy = ai.reverse_sbs_score(x, y)
labels = Preprocessing.get_labels()
print(f"Max accuracy is {accuracy}")
print("Features:")
for f in features:
print(labels[f])
def get_hasil_uji():
print('start...')
preprocessing = Preprocessing()
dataset = preprocessing.read_ringkas()
hasil_sistem = preprocessing.read_hasil()
kalimat_ringkas_dataset = preprocessing.split_sentence(dataset)
kalimat_ringkas_hasil = preprocessing.split_sentence(hasil_sistem)
akurasi = []
hasil = []
for i in range(len(dataset)):
fn = 0
fp = 0
tp = len(kalimat_ringkas_dataset[i])
for sentence in kalimat_ringkas_dataset[i]:
if not (sentence in kalimat_ringkas_hasil[i]):
fn += 1
for sentence in kalimat_ringkas_hasil[i]:
if not (sentence in kalimat_ringkas_dataset[i]):
fp += 1
precission = tp / (tp + fp)
recall = tp / (tp + fn)
f1 = 2 * ((precission * recall) / (precission + recall))
hasil.append({
"akurasi": f1,
"system": hasil_sistem[i],
"pakar": dataset[i],
"precission": precission,
"recall": recall
})
akurasi.append(f1)
print(np.average(akurasi))
print(np.max(akurasi))
print(np.min(akurasi))
return jsonify(hasil), 200
def _random_forest_scores(self):
print("Looking for best features...")
x, y = self._get_data_tuples()
ai = AI()
ai.random_forest_score(x, y, Preprocessing.get_labels())
class Predictor:
"""
This class is used to predict emotions in practice
"""
def __init__(self, freq, model):
self.freq = freq
self.model = model
self.pre = Preprocessing(self.freq)
self.post = Postprocessing(self.freq)
self.base_bvp_features = None
self.base_gsr_features = None
def process(self):
"""
Loop that predicts emotions. After small changes may be used to get signals via pipeline
"""
base_bvp, base_gsr = self._get_basic_values()
print("Processing base signals...")
self.base_bvp_features = self.pre.get_base_bvp_features(base_bvp)
self.base_gsr_features = self.pre.get_base_gsr_features(base_gsr)
while True:
print(
"Enter BVP signal (one value per line). Press ENTER to finish: "
)
bvp = self._read_signal()
print(
"Enter GSR signal (one value per line). Press ENTER to finish: "
)
gsr = self._read_signal()
values = self.pre.get_diffed_values(bvp, gsr,
self.base_bvp_features,
self.base_gsr_features)
values = self.post.standarize(values)
emotion = self.model.predict(values)
print(f"Predicted emotion class: {emotion}")
def _read_signal(self):
signal = []
for line in sys.stdin:
rawline = line.rstrip()
if not rawline:
break
signal.append(float(rawline))
return signal
def _get_basic_values(self):
print(
"Enter base BVP signal (one value per line). Press ENTER to finish: "
)
base_bvp = self._read_signal()
print(
"Enter base GSR signal (one value per line). Press ENTER to finish: "
)
base_gsr = self._read_signal()
return base_bvp, base_gsr