【数据分析&数据挖掘】数据合并和拼接案例

KaiZhaoKZ 2019-12-29

import pandas as pd
import numpy as np

# 加载数据——detail
detail_1 = pd.read_excel("./meal_order_detail.xlsx", sheetname=0)
detail_2 = pd.read_excel("./meal_order_detail.xlsx", sheetname=1)
detail_3 = pd.read_excel("./meal_order_detail.xlsx", sheetname=2)

print("detail_1 的形状: \n", detail_1.shape)
print("detail_1 的列索引: \n", detail_1.columns)
print("detail_2 的形状: \n", detail_2.shape)
print("detail_2 的列索引: \n", detail_2.columns)
print("detail_3 的形状: \n", detail_3.shape)
print("detail_3 的列索引: \n", detail_3.columns)

print("~"*60)
# 将detail_2, detail_3直接追加到detaiL_1下面
detail = pd.concat((detail_1, detail_2, detail_3), axis=0, join="inner")
print("detail的形状;\n", detail.shape)


# 加载info
info = pd.read_csv("./meal_order_info.csv", encoding="ansi")
print("info: \n", info.shape)

# info与detail进行主键拼接
res = pd.merge(left=detail, right=info, left_on="order_id", right_on="info_id", how="inner")
res = pd.merge(left=detail, right=info, left_on="order_id", right_on="info_id", how="left")
print("info与detail主键拼接的结果为: \n", res.shape)
print("res的列名: \n", res.columns)


# 加载users
users = pd.read_excel("./users.xlsx")
# info与detail进行主键拼接的结果与users进行主键拼接
res = pd.merge(left=res, right=users, left_on="name", right_on="ACCOUNT", how="inner")
print("最终进行主键拼接的结果: \n", res)
print("最终res的列名称: \n", res.columns)

print("name与ACCOUNT对比相同", np.all(res.loc[:, "name"] == res.loc[:, "ACCOUNT"]))
print("order_id与info_id对比相同", np.all(res.loc[:, "order_id"] == res.loc[:, "info_id"]))
print("emp_id_x与emp_id_y对比相同", np.all(res.loc[:, "emp_id_x"] == res.loc[:, "emp_id_y"]))

res.drop(labels=["ACCOUNT", "info_id", "emp_id_y"], axis=1, inplace=True)

print("删除3列之后的结果: \n", res.shape)
print("删除3列之后的结果: \n", res.columns)

drop_list = []
for column in res.columns:
    # 统计每一列的非空数据的数量
    res_count = res.loc[:, column].count()
    # 如果整列非空数据的数量为0,意味着整列都是空的
    if res_count == 0:
        drop_list.append(column)

# 删除整列为空的列
res.drop(labels=drop_list, axis=1, inplace=True)
print("去除整列为空的数据之后的结果: \n", res.shape)
print("去除整列为空的数据之后的结果: \n", res.columns)

drop_dup_list = []
# 如果整列数据完全相同——该列, 该属性对于区分各列没有意义
for column in res.columns:
    res_ = res.drop_duplicates(subset=column, inplace=False)
    if res_.shape[0] == 1:
        print("res_.shape[0]: \n", res_.shape[0])
        drop_dup_list.append(column)

# 删除全部一样的列
res.drop(labels=drop_dup_list, axis=1, inplace=True)
print("最终的结果: \n", res.shape)
print("最终的结果: \n", res.columns)

相关推荐