电商数据可视化分析

# -*- encoding: utf-8 -*-
"""
@File    : homework_6_电商数据可视化分析.py
@Time    : 2019/9/24 10:31
@Author  : chen


"""
# 可以打开下面的网页，看各种图形的源码
# https://plot.ly/python/


import pandas as pd
import numpy as np

import plotly.graph_objects as go
# pip install colorlover
import colorlover as cl

colors = ['#F1948A', '#AED6F1', '#F9E79F', '#E5E8E8', '#F1948A', '#D0ECE7', '#F6DDCC', '#D2B4DE',
          '#117A65', '#FAE5D3', '#34495E', '#DC7633', '#D35400', '#0E6251', '#FCF3CF', '#E8F8F5', '#D4E6F1', '#FAD8D8', '#E59866']

data = pd.read_csv('BlackFriday.csv')
print(data.head(10))     # 前10条数据

print(data.shape)        # 数据维度
print(data.Occupation.unique())          #
print(len(data.Occupation.unique()))     # 数据长度

print(data.describe())    # 输出数据

# 缺失值查看
print(data.isna().sum())

# pivot_table  Pandas的高级应用中的透视表的功能
gender_purchase = data.pivot_table(values='Purchase', aggfunc="sum", index=["User_ID","Gender"]).reset_index()    # aggfunc="sum"代表指定的函数  ["User_ID","Gender"]按照ID，性别分组
gender_purchase.head(20)   # 前20个数据
print(gender_purchase.count())


gender_count = gender_purchase.groupby(by="Gender").size().reset_index(name="人数")
gender_count["占比"] = gender_count["人数"]/gender_count["人数"].sum()
print(gender_count)

# 描绘数据
trace = go.Pie(labels=gender_purchase.Gender.tolist(),
                values=gender_purchase.Purchase.tolist(),
                hole=0.5)
fig= go.Figure(data=[trace])
fig.show()

# 购物分布   男女性购物 箱型图
x_female = gender_purchase[gender_purchase.Gender == "F"].Purchase
y_male = gender_purchase[gender_purchase.Gender == "M"].Purchase
trance1= go.Box(y=y_male, name="男性购物", boxmean=True)                  # y=y_male 改为x=y_male  可以图像横着显示
trance2= go.Box(y=x_female, name="女性购物", boxmean=True)
fig = go.Figure(data=[trance1, trance2])
fig.show()

# 销售前10
top10_sellers = data.pivot_table(values=['Purchase'],
                                 index=['Product_ID'],
                                 aggfunc='count').reset_index().sort_values(by='Purchase',ascending=False).head(10)
print("top10_sellers:", top10_sellers)

# 购买热销产品的是谁
top_sellers_buyers = data[data.Product_ID.isin(top10_sellers.Product_ID.tolist())]
print(top_sellers_buyers.head(10))

# 热销商品和性别的关系
top_sellers_gender = top_sellers_buyers.pivot_table(values="Purchase",
                                                    index=["Product_ID","Gender"],
                                                    aggfunc="count").reset_index()
print(top_sellers_gender)

traces=[]
i=0
for g in top_sellers_gender.Gender.unique():
    trace = go.Bar(x=top_sellers_gender[top_sellers_gender.Gender==g].Purchase,
                   y=top_sellers_gender[top_sellers_gender.Gender==g].Product_ID,
                   name=g,                                  # 横轴的名称
                   marker= dict(color=colors[i]),           # 颜色
                   orientation = "h")                      # 方向控制
    traces.append(trace)
    i+=1
fig = go.Figure(data=traces)
fig.show()

# 热销产品和城市之间的关系
top_sellers_city = top_sellers_buyers.prvot_table(values="Purchase",
                                                  index=['Product_ID',"City_Category"],
                                                  aggfunc="count").reset_index()
print(top_sellers_city.head())
traces=[]
i=0
for c in top_sellers_city.City_Category.unique():
    trace = go.Bar(x=top_sellers_city[top_sellers_city.City_Category==c].Purchase,
                   y=top_sellers_city[top_sellers_city.City_Category==c].Product_ID,
                   name=c,  # 横轴的名称
                   marker=dict(color=colors[i]),  # 颜色
                   orientation="h")  # 方向控制
    traces.append(trace)
    i += 1
go.Figure(data=traces).show()