def test_write_data(self):
data_path_test = "./data/data0"
write_data(self.data, data_path_test)
with open(data_path_test, "r") as f:
lines = f.readlines()
self.assertEqual(len(lines), len(self.data))
self.assertEqual(json.loads(lines[0]), self.data[0])
self.assertEqual(json.loads(lines[-1]), self.data[-1])
os.remove(data_path_test)
print(num_dic)
train_num_dic = {}
valid_num_dic = {}
for i in num_dic:
train_num_dic[i] = int(train_ratio * num_dic[i])
valid_num_dic[i] = int(valid_ratio * num_dic[i])
random.seed(30)
random.shuffle(data)
for idx, item in enumerate(data):
for i in num_dic:
if item["gesture"] == i:
if train_num_dic[i] > 0:
train_data.append(item)
train_num_dic[i] -= 1
elif valid_num_dic[i] > 0:
valid_data.append(item)
valid_num_dic[i] -= 1
else:
test_data.append(item)
print("train_length:" + str(len(train_data)))
print("test_length:" + str(len(test_data)))
return train_data, valid_data, test_data
if __name__ == "__main__":
data = read_data("./data/complete_data")
train_data, valid_data, test_data = split_data(data, 0.6, 0.2)
write_data(train_data, "./data/train")
write_data(valid_data, "./data/valid")
write_data(test_data, "./data/test")