# Import data as pandas dataframe
if test_data_gen == 'False' or test_data_gen == 'false':
print("Parsing data...")
print("\n")
data = pdp.pre_vec_parser(filepath, window_size)
# Vectorise data
print("Vectorising data...")
print("\n")
if use_pssm == False:
X_test, Y_test = ddp.skl_parser(data, window_size)
else:
pssm_loc = repo_loc + 'SignalP/input/test_pssms/'
data_pssm = pdp.pssm_parser(data, window_size, pssm_loc)
X_test, Y_test = pdp.skl_pssm_parser(data,
window_size,
pssm_type='freq')
# Test model
scores = oD()
scores['labels'] = np.array([-1, 0, 1])
model = joblib.load(inpath + model_name)
# Import data as pandas dataframe
print("Parsing data...")
print("\n")
data = pdp.pre_vec_parser(filepath, window_size)
# Vectorise data
print("Vectorising data...")
print("\n")
if use_pssm == False:
X, Y = ddp.skl_parser(data, window_size)
else:
data_pssm = pdp.pssm_parser(data, window_size, pssm_loc)
X, Y = pdp.skl_pssm_parser(data_pssm, window_size, pssm_type=pssm_type)
# Train model
model = RandomForestClassifier(n_estimators = n_estimators, class_weight='balanced')
print("Training this magnificient model...")
print("\n")
fit_start = time.time()
scores = oD()
scores['labels'] = np.array([-1,0,1])
# Starting script
start = time.time()
f1_scorer = make_scorer(f1_score, labels=[-1,1], average='macro')
final_list = oD()
for windows in window_size:
data = ddp.pre_vec_parser(filepath, windows)
clf = LinearSVC(class_weight = 'balanced')
X, Y = ddp.skl_parser(data, windows)
parameters = {"C": [1,2,4,8] }
model_tunning = GridSearchCV(clf, param_grid=parameters, scoring=f1_scorer, n_jobs=-3)
model_tunning.fit(X,Y)
s = model_tunning.best_score_
p = model_tunning.best_params_
final_list[str(windows)] = [s,p['C']]
end = time.time()
best_table = pd.DataFrame.from_dict(final_list, orient='index')
clf = LinearSVC(class_weight='balanced')
scores = oD()
scores['labels'] = np.array([-1, 0, 1])
p = np.zeros(3)
r = np.zeros(3)
f = np.zeros(3)
s = np.zeros(3)
ft = np.zeros(3)
for train_data, test_data in ddp.cv_data_gen(data,
cv_sets,
randomise=False):
X_train, Y_train = ddp.skl_parser(train_data, windows)
X_test, Y_test = ddp.skl_parser(test_data, windows)
fit_start = time.time()
clf.fit(X_train, Y_train)
fit_end = time.time()
fit_time = fit_end - fit_start
predicted = clf.predict(X_test)
precision, recall, fscore, support = score(Y_test, predicted)
p = p + precision
r = r + recall
f = f + fscore