def predict_algorithm(_flavor_name, _ecs_array, _predict_time, _alpha, _beta):
predict_result = []
# 获取日期,频数对
date_freq_dict = input_mat.get_date_freq_pair(_flavor_name, _ecs_array,
_predict_time)
print _flavor_name, " date_freq_dict：\n", date_freq_dict
# 获取 输入
X, y, train_data_nums = input_mat.get_X_y(date_freq_dict)
print _flavor_name, "X,y: ", X, y
# 获取 模型
reg_obj = Regression(X, y) #_alpha
# 预测
predict_date_list = input_mat.get_predicting_date_list(_predict_time)
for date in predict_date_list:
predict_X = input_mat.get_predict_X(date, date_freq_dict,
train_data_nums)
predict_X = [predict_X]
predict_num_a_day = reg_obj.predict_ridge(
predict_X) #predict_kernal_ridge(predict_X)#
if reg_obj.isError == True:
sum_p = 0
for item in predict_X[0]:
sum_p += item
predict_num_a_day = sum_p / len(predict_X[0])
else:
predict_num_a_day = predict_num_a_day[0][0]
predict_result.append(predict_num_a_day)
input_mat.add_date_freq_dict(date, predict_num_a_day,
date_freq_dict) # predict_num_a_day[0][0]
#print "date_freq_dict",date_freq_dict
print '\n' + _flavor_name + ' predict_result: ', predict_result
# 对预测的每天数量累加
predict_flavors_sum = 0 # add---
for flavor in predict_result:
# predict_flavors_sum += max(float(flavor[0][0]),0)
predict_flavors_sum += float(flavor) # flavor[0][0]
predict_flavors_sum = max(predict_flavors_sum, 0)
return math.ceil(predict_flavors_sum)
