python数据分析案例1

最新推荐文章于 2023-03-30 09:11:14 发布

y_j_y_

最新推荐文章于 2023-03-30 09:11:14 发布

阅读量776

点赞数

分类专栏：数据分析回归文章标签： python数据分析

本文链接：https://blog.csdn.net/y_j_y_/article/details/79235455

版权

数据分析同时被 2 个专栏收录

2 篇文章 0 订阅

订阅专栏

回归

1 篇文章 0 订阅

订阅专栏

import pandas as pd
import matplotlib.pyplot as plt

dir = './data/'
train = pd.read_table(dir + 'train_20171215.txt',engine='python')
test_A = pd.read_table(dir + 'test_A_20171225.txt',engine='python')
sample_A = pd.read_table(dir + 'sample_A_20171225.txt',engine='python',header=None)
sample_A.columns = ['date','day_of_week']

# 因为第一赛季只是预测与时间相关的cnt的数量
# 所以可以对数据以date和day_of_week进行数据合并
train = train.groupby(['date','day_of_week'],as_index=False).cnt.sum()
# print(train)
# plt.plot(train['day_of_week'],train['cnt'],'*')
# plt.show()

#观察星期约束下销量在时间轴上的分布图
# for i in range(7):
#     tmp = train[train['day_of_week']==i+1]
#     plt.subplot(7, 1, i+1)
#     plt.plot(tmp['date'],tmp['cnt'],'*')
# plt.show()

#筛选测试集和训练集
xx_train = train[train['date']<=756]
xx_test = train[train['date']>756]
print('test shape',xx_test.shape)
print('train shape',xx_train.shape)

# 方案零：均值大法（原始数据验证）
from sklearn.metrics import mean_squared_error
# 线下统计每周的均值数据，不加权
xx_train = xx_train.groupby(['day_of_week'],as_index=False).cnt.mean()
xx_result = pd.merge(xx_test,xx_train,on=['day_of_week'],how='left')
print('xx_result shape',xx_result.shape)
print(xx_result)
print(mean_squared_error(xx_result['cnt_x'],xx_result['cnt_y']))