python分析电商交易数据 源码+数据

import pandas as pd import matplotlib.pyplot as plt def address_chart(df): add = df.groupby('address')['id'].count().sort_values(ascending=False) plt.rcParams['font.sans-serif'] = ['SimHei'] add.plot.pie(figsize=(12,6),labels=add.index, autopct='%1.1f') plt.title('买家地区分布') plt.show() def time_chart(df): df1 = df.copy() df1['paytime'] = df1['paytime'].dt.time df1['paytime'] = pd.to_datetime(df.paytime) plt.figure(figsize=(20,8), dpi=80) s = df1['paytime'].dt.floor('30T') df1['paytime'] = s.dt.strftime('%H:%M') + '-' + (s+pd.Timedelta(29*60,unit='s')).dt.strftime("%H:%M") timedf1 = df1.groupby('paytime')['id'].count() timedf1.drop(index='NaT-NaT', inplace=True) timedf_x = timedf1.index timedf_y = timedf1.values plt.xticks(rotation=60) plt.plot(timedf_x, timedf_y) plt.show() def money_chart(df): df['day'] = df['ordertime'].dt.day xx = df.groupby('day')['amount'].sum().index yy_1 = df.groupby('day')['amount'].sum().values/10000 yy_2 = df.groupby('day')['paid'].sum().values/10000 yy_3 = df.groupby('day')['refund'].sum().values/10000 plt.plot(xx,yy_1,label='Order Amount',color='red',marker='+') plt.plot(xx,yy_2,label='Payment Amount',color='green',marker='o') plt.plot(xx,yy_3,label='Refund Amount',color='blue',marker='.') plt.xlabel('day') plt.ylabel('money(10000yuan)') plt.show() def taobao_analysis(csv_path): df = pd.DataFrame(pd.read_csv(csv_path)) # id 总金额 实际支付金额 地址 订单创建时间 付款时间 退款金额 df.columns = ['id', 'amount', 'paid', 'address', 'ordertime', 'paytime', 'refund'] df.paytime = pd.to_datetime(df.paytime) df.ordertime = pd.to_datetime(df.ordertime) # 订单交易数量 order_num = df.id.count() # 退货订单数量 refund_num = df[df['refund'] > 0].refund.count() # 退货率 refund_rate = round(refund_num / order_num * 100, 3) # 交易总金额 amount_sum = df.amount.sum() # 成交总金额 paid_sum = df.paid.sum() # 退款总金额 refund_sum = df.refund.sum() # 实际成交额 paymey = paid_sum - refund_sum # 买家地区饼图 # address_chart(df) # 支付金额时间分布 # time_chart(df) # 销售额走势 money_chart(df) print('订单数量：', order_num) print('退货订单数量：', refund_num) print('退货率：{}%'.format(refund_rate)) print('交易总金额：', amount_sum) print('成交总金额：', paid_sum) print('退款总金额：', refund_sum) print('实际成交额：', paymey) if __name__ == '__main__': csv_path = r'./data/tmall_order_report.csv' taobao_analysis(csv_path)