import pandas as pd
import numpy as np
from pyecharts import Line
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LinearRegression, SGDRegressor

pd.set_option('expand_frame_repr', False)


def bubble_sort(nums, nums1):
    for i in range(len(nums) - 1):  # 这个循环负责设置冒泡排序进行的次数
        for j in range(len(nums) - i - 1):  # ｊ为列表下标
            if nums[j] > nums[j + 1]:
                nums[j], nums[j + 1] = nums[j + 1], nums[j]
                nums1[j], nums1[j + 1] = nums1[j + 1], nums1[j]
    return nums, nums1


def parse():
    # 读取数据
    data = pd.read_csv('./bk_esf.csv')
    # print(data)
    # print('---------------')

    # 处理有无电梯　有－2，无－1
    data['elevator'] = data['elevator'].replace('有', 2)
    data['elevator'] = data['elevator'].fillna(1)

    # 产权　转化成数值类型
    data['property'] = data['property'].replace('未知', np.nan)
    data['property'] = data['property'].replace('[^\d]', '', regex=True)
    data['property'] = pd.to_numeric(data['property'])
    data['property'] = data['property'].fillna(data['property'].mean())

    # # 房屋结构　暂无数据－1，平层－2，复式－3,
    # data['house_category'] = data['house_category'].replace('暂无数据', 1)
    # data['house_category'] = data['house_category'].replace('平层', 2)
    # data['house_category'] = data['house_category'].replace('复式', 3)

    # 建筑年代 转化成数值类型
    data['building_age'] = data['building_age'].replace('[^\d]+', '', regex=True)
    data['building_age'] = pd.to_numeric(data['building_age'])
    data['building_age'] = data['building_age'].fillna(data['building_age'].mean())

    # 总楼层，转化成数值
    data['floor_count'] = data['floor_count'].replace('[^\d]+', '', regex=True)
    data['floor_count'] = pd.to_numeric(data['floor_count'])
    data['floor_count'] = data['floor_count'].fillna(data['floor_count'].mean())

    # # 楼层，低－1，中－2，高－3，地下室－4
    # data['floor'] = data['floor'].replace('低楼层', 1)
    # data['floor'] = data['floor'].replace('中楼层', 2)
    # data['floor'] = data['floor'].replace('高楼层', 3)
    # data['floor'] = data['floor'].replace('地下室', 4)

    # 户型  2室1厅-转化为21　3室1厅－转化为31
    data['house_type'] = data['house_type'].replace('室|厅', '', regex=True)
    data['house_type'] = pd.to_numeric(data['house_type'])

    # 删除不影响目标值的特征
    data = data.drop(['category', 'title', 'unit', 'average', 'community', 'status', 'crawl_time'], axis=1)

    # 拿到目标值
    target = data['amount']
    data = data.drop(['amount'], axis=1)
    # data.insert(data.shape[1], 'amount', target)

    print(data, data.shape)
    print('---------------------')

    # one hot编码
    data = pd.get_dummies(data)

    print(data, data.shape)

    x_train, x_test, y_train, y_test = train_test_split(data, target, test_size=0.25, random_state=1)

    # 标准化
    std_x = StandardScaler()
    std_y = StandardScaler()

    # 标准化ｘ
    x_train = std_x.fit_transform(x_train)
    x_test = std_x.transform(x_test)

    # 标准化ｙ
    y_train = std_y.fit_transform(y_train.values.reshape(-1, 1))
    y_test = std_y.transform(y_test.values.reshape(-1, 1))

    # 正规方程
    # lr = LinearRegression()
    # 梯度下降
    sgd = SGDRegressor(random_state=1)

    # lr.fit(x_train, y_train)
    sgd.fit(x_train, y_train)

    # print('参数权重:', lr.coef_)
    print('参数权重:', sgd.coef_)

    # y_predict = std_y.inverse_transform(lr.predict(x_test))
    y_predict = std_y.inverse_transform(sgd.predict(x_test))

    print('预测结果:', y_predict)
    xx = std_x.transform(data[:1])
    yy = std_y.inverse_transform(sgd.predict(xx))
    print(yy)

    # 绘制折线图展示
    # x_axis = list(range(1, y_test.shape[0] + 1))
    # line = Line("线性回归测试集预测结果可视化", width=1200)
    # # y_test_s, y_pred_s = bubble_sort(y_test, y_pred)
    # line.add("真实值", x_axis, std_y.inverse_transform(y_test), is_smooth=True)
    # line.add("预测值", x_axis, y_predict, is_smooth=True)
    # line.render(path='./echart_html/房价回归测试集预测结果可视化.html')


if __name__ == '__main__':
    parse()