def prediction(main_sym, add_sym, ton, pul):
ncluster = 10
nsample = 500
feature_dict, tag_dict = cnn.create_dict('./app/algo/data/')
vec = cnn.create_vec(main_sym, add_sym, ton, pul, feature_dict)
workbook = cnn.xlrd.open_workbook('./app/algo/data/data.xlsx')
sheet = workbook.sheet_by_index(1)
rows = sheet.nrows
raw = []
for i in range(1, rows):
row = []
for j in range(5):
row.append(sheet.row_values(i)[j])
raw.append(row)
X, y = cnn.create_matrix(raw, feature_dict, tag_dict)
X = X[0:nsample]
X.append(vec)
X = construct_W(np.array(X))
y_pred = SpectralClustering(n_clusters=ncluster,
gamma=1,
affinity='precomputed',
n_jobs=1).fit_predict(X)
cluster = []
for i in range(nsample):
if y_pred[i] == y_pred[-1]:
cluster.append(y[i])
count = np.sum(cluster, axis=0)
thre = np.max(count) * 0.75
res = []
for i in range(len(count)):
if count[i] >= thre:
res.append(list(tag_dict.keys())[list(tag_dict.values()).index(i)])
return res
def train():
ncluster = 10
nsample = 500
feature_dict, tag_dict = cnn.create_dict('./data/')
workbook = cnn.xlrd.open_workbook('data/data.xlsx')
sheet = workbook.sheet_by_index(1)
rows = sheet.nrows
raw = []
for i in range(1, rows):
row = []
for j in range(5):
row.append(sheet.row_values(i)[j])
# print(row)
raw.append(row)
X, y = cnn.create_matrix(raw, feature_dict, tag_dict)
X = np.array(X[0:nsample])
X = construct_W(X)
y_pred = SpectralClustering(n_clusters=ncluster,
gamma=1,
affinity='nearest_neighbors').fit_predict(X)
clusters = [[] for i in range(ncluster)]
tags = []
for i in range(nsample):
clusters[y_pred[i]].append(y[i])
for i in clusters:
count = np.sum(i, axis=0)
thre = np.max(count) * 0.75
tag = [1 if j > thre else 0 for j in count]
tags.append(tag)
y_ = [tags[y_pred[i]] for i in range(nsample)]
h, o, c = cnn.assess(y_, y[0:nsample])
print('Hamming Loss:' + str(h) + '\t One-Error:' + str(o) +
'\t Coverage:' + str(c))
def prediction(main_sym, add_sym, ton, pul):
feature_dict, tag_dict = cnn.create_dict('./app/algo/data/')
workbook = cnn.xlrd.open_workbook('./app/algo/data/data.xlsx')
sheet = workbook.sheet_by_index(1)
rows = sheet.nrows
raw = []
for i in range(1, rows):
row = []
for j in range(5):
row.append(sheet.row_values(i)[j])
raw.append(row)
X, y_ = cnn.create_matrix(raw, feature_dict, tag_dict)
y = [i.index(1) for i in y_]
classifier = AdaBoostClassifier()
classifier.fit(X, y)
feature_dict, tag_dict = cnn.create_dict('./app/algo/data/')
vec = cnn.create_vec(main_sym, add_sym, ton, pul, feature_dict)
y_ = classifier.predict([vec])[0]
return [list(tag_dict.keys())[list(tag_dict.values()).index(y_)]]
def train():
feature_dict, tag_dict = cnn.create_dict('./data/')
workbook = cnn.xlrd.open_workbook('data/data.xlsx')
sheet = workbook.sheet_by_index(1)
rows = sheet.nrows
raw = []
for i in range(1, rows):
row = []
for j in range(5):
row.append(sheet.row_values(i)[j])
raw.append(row)
X, y = cnn.create_matrix(raw, feature_dict, tag_dict)
classifier = RandomForestClassifier()
classifier.fit(X, y)
y_ = classifier.predict(X)
h, o, c = cnn.assess(y_, y)
print('Hamming Loss:' + str(h) + '\t One-Error:' + str(o) +
'\t Coverage:' + str(c))
def train():
feature_dict, tag_dict = cnn.create_dict('./data/')
workbook = cnn.xlrd.open_workbook('data/data.xlsx')
sheet = workbook.sheet_by_index(1)
rows = sheet.nrows
raw = []
for i in range(1, rows):
row = []
for j in range(5):
row.append(sheet.row_values(i)[j])
raw.append(row)
X, y_ = cnn.create_matrix(raw, feature_dict, tag_dict)
y = [i.index(1) for i in y_]
classifier = AdaBoostClassifier()
classifier.fit(X, y)
y_ = classifier.predict(X)
accuracy, micro_precise, micro_recall, micro_f1, macro_precise, macro_recall, macro_f1 = assess(
y_, y)
print('Accuracy:' + str(accuracy) + '\t Micro_Precise:' +
str(micro_precise) + '\t Micro_Recall:' + str(micro_recall) +
'\t Micro_F1:' + str(micro_f1) + '\t Macro_Precise:' +
str(macro_precise) + '\t Macro_Recall:' + str(macro_recall) +
'\t Macro_F1:' + str(macro_f1))