? 作者：韩信子@ShowMeAI
? 数据分析实战系列：https://www.showmeai.tech/tutorials/40
? 本文地址：https://www.showmeai.tech/article-detail/388
? 声明：版权所有，转载请联系平台与作者并注明出处
? 收藏ShowMeAI查看更多精彩内容

在本篇内容中，ShowMeAI将带大家对旅游业，主要是酒店预订需求进行分析，我们使用到的数据集包含城市酒店和度假酒店的预订信息，包括预订时间、住宿时长、客人入住的周末或工作日晚数以及可用停车位数量等信息。

我们本次用到的是 ?酒店预订数据集，包含 119390 位客人，有 32 个特征字段，大家可以通过 ShowMeAI 的百度网盘地址下载。

? 实战数据集下载（百度网盘）：公✦众✦号『ShowMeAI研究中心』回复『实战』，或者点击 这里 获取本文 [59]旅游业大数据多维度业务分析案例 『酒店预订数据集』

⭐ ShowMeAI官方GitHub：https://github.com/ShowMeAI-Hub

本文数据分析部分涉及的工具库，大家可以参考ShowMeAI制作的工具库速查表和教程进行学习和快速使用。

? 数据科学工具库速查表 | Pandas 速查表
? 图解数据分析：从入门到精通系列教程

? 导入工具库

# 数据处理&科学计算import pandas as pdimport numpy as np# 数据分析&绘图import matplotlib.pyplot as pltimport seaborn as snsimport plotly.express as pximport plotly.graph_objects as goimport plotly.figure_factory as ffimport warningswarnings.filterwarnings("ignore")# 科学计算from scipy.stats import skew,kurtosisimport pylab as py# 时间import timeimport datetimefrom datetime import datetimefrom datetime import date

? 读取数据

df = pd.read_csv("hotel_bookings.csv")df.head()

? 数据信息初览

df.info()

? 数据预处理? 清洗&缺失值处理

首先统计字段缺失值比例

df.isnull().sum().sort_values(ascending = False) / len(df)

我们对有缺失的字段做一些缺失值填充工作

# 填充"agent" 和 "company" 字段中的缺失值df["agent"].fillna(0, inplace = True)df["company"].fillna(0, inplace = True)# 使用众数填充"country"字段缺失值df["country"].fillna(df["country"].mode()[0], inplace = True)# 删除包含"children"缺失值的数据记录df.dropna(subset = ["children"], axis = 0, inplace = True)

? 字段数据处理

# 将“distribution_channel”列中的“Undefined”转换为“TA/TO”df["distribution_channel"].replace("Undefined", "TA/TO", inplace = True)# meal字段映射处理df["meal"].replace(["Undefined", "BB", "FB", "HB", "SC" ], ["No Meal", "Breakfast", "Full Board", "Half Board", "No Meal"], inplace = True)# 将“is_canceled”列的值从 0 和 1 转换为“Cancelled”和“Not Cancelled”df["is_canceled"].replace([0, 1], ["Cancelled", "Not Cancelled"], inplace = True)

? 调整数据类型

  • 将children、agent和company列的数据类型转换为整型
  • 将reservation_status_date列的数据类型从对象转换为日期类型
# 转整型df["children"].astype(int)df["agent"].astype(int)df["company"].astype(int)# 时间型pd.to_datetime(df["reservation_status_date"])

? 重复数据处理

df.drop_duplicates(inplace = True)

? 构建汇总字段

我们对顾客总体的居住晚数进行统计

df["total_nights"] = df["stays_in_weekend_nights"] + df["stays_in_week_nights"]

? 描述性统计

我们基于pandas的简单功能，对数据的统计分布做一个处理了解

df.describe().T

? 探索性数据分析? 酒店维度分析

# 我们对 城市酒店 和 度假酒店 进行统计分析labels = ['City Hotel', 'Resort Hotel']colors = ["#538B8B", "#7AC5CD"]order = df['hotel'].value_counts().indexplt.figure(figsize = (19, 9))plt.suptitle('Bookings By Hotels', fontweight = 'heavy', fontsize = '16',            fontfamily = 'sans-serif', color = "black")# Pie Chartplt.subplot(1, 2, 1)plt.title('Pie Chart', fontweight = 'bold', fontfamily = "sans-serif", color = 'black')plt.pie(df["hotel"].value_counts(), pctdistance = 0.7, autopct = '%.2f%%', labels = labels,wedgeprops = dict(alpha = 0.8, edgecolor = "black"), textprops = {'fontsize': 12}, colors = colors)centre = plt.Circle((0,0), 0.45, fc = "white", edgecolor = "black")plt.gcf().gca().add_artist(centre)# Histogramcountplt = plt.subplot(1, 2, 2)plt.title("Histogram", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "hotel", data = df, order = order, edgecolor = "black", palette = colors)for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Hotel", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.xticks([0, 1], labels)plt.grid(axis = "y", alpha = 0.4)df['hotel'].value_counts()

? 结论：超过 60% 的预订酒店是城市酒店

? 细分市场分析

labels = ["Online TA", "Offline TA/TO", "Direct", "Groups", "Corporate", "Complementary", "Aviation"]order = df['market_segment'].value_counts().indexplt.figure(figsize = (22, 9))plt.suptitle('Bookings By Market Segment', fontweight = 'heavy', fontsize = '16',            fontfamily = 'sans-serif', color = "black")# Pie Chartplt.subplot(1, 2, 1)plt.title('Pie Chart', fontweight = 'bold', fontfamily = "sans-serif", color = 'black')plt.pie(df["market_segment"].value_counts(), pctdistance = 0.7, autopct = '%.2f%%', labels = labels,wedgeprops = dict(alpha = 0.8, edgecolor = "black"), textprops = {'fontsize': 12})centre = plt.Circle((0,0), 0.45, fc = "white", edgecolor = "black")plt.gcf().gca().add_artist(centre)# Histogramcountplt = plt.subplot(1, 2, 2)plt.title("Histogram", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "market_segment", data = df, order = order, edgecolor = "black",)for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Market Segment", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df['market_segment'].value_counts()

? 结论：超过 50% 的预订是通过在线旅行社完成的。

? 分销渠道分析

colors = ["#8B7D6B", "#000000", "#CDB79E", "#FFE4C4"]labels = ["TA/TO", "Direct", "Corporate", "GDS"]order = df['distribution_channel'].value_counts().indexplt.figure(figsize = (19, 9))plt.suptitle('Bookings By Distribution Channel', fontweight = 'heavy', fontsize = '16',            fontfamily = 'sans-serif', color = "black")# Pie Chartplt.subplot(1, 2, 1)plt.title('Pie Chart', fontweight = 'bold', fontfamily = "sans-serif", color = 'black')plt.pie(df["distribution_channel"].value_counts(), pctdistance = 0.7, autopct = '%.2f%%', labels = labels, colors = colors,wedgeprops = dict(alpha = 0.8, edgecolor = "black"), textprops = {'fontsize': 12})centre = plt.Circle((0,0), 0.45, fc = "white", edgecolor = "black")plt.gcf().gca().add_artist(centre)# Histogramcountplt = plt.subplot(1, 2, 2)plt.title("Histogram", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "distribution_channel", data = df, order = order, edgecolor = "black", palette = colors)for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Distribution Channel", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df['distribution_channel'].value_counts()

? 结论：超过 80% 的预订是通过旅行社/运营商完成的。

? 餐食分析

colors = ["#6495ED", "#1874CD", "#009ACD", "#00688B"]labels = ["Breakfast", "No Meal", "Half Board", "Full Board"]order = df['meal'].value_counts().indexplt.figure(figsize = (19, 9))plt.suptitle('Bookings By Meals', fontweight = 'heavy', fontsize = '16',            fontfamily = 'sans-serif', color = "black")# Pie Chartplt.subplot(1, 2, 1)plt.title('Pie Chart', fontweight = 'bold', fontfamily = "sans-serif", color = 'black')plt.pie(df["meal"].value_counts(), pctdistance = 0.7, autopct = '%.2f%%', labels = labels, colors = colors,wedgeprops = dict(alpha = 0.8, edgecolor = "black"), textprops = {'fontsize': 12})centre = plt.Circle((0,0), 0.45, fc = "white", edgecolor = "black")plt.gcf().gca().add_artist(centre)# Histogramcountplt = plt.subplot(1, 2, 2)plt.title("Histogram", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "meal", data = df, order = order, edgecolor = "black", palette = colors)for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Meal", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df['meal'].value_counts()

? 结论：超过 70% 的客人预订早餐，近 90% 的客人预订餐点。

? 顾客类型分析

labels = ["Transient", "Transient-Party", "Contract", "Group"]order = df['customer_type'].value_counts().indexplt.figure(figsize = (19, 9))plt.suptitle('Bookings By Customer Type', fontweight = 'heavy', fontsize = '16',            fontfamily = 'sans-serif', color = "black")# Pie Chartplt.subplot(1, 2, 1)plt.title('Pie Chart', fontweight = 'bold', fontfamily = "sans-serif", color = 'black')plt.pie(df["customer_type"].value_counts(), pctdistance = 0.7, autopct = '%.2f%%', labels = labels,wedgeprops = dict(alpha = 0.8, edgecolor = "black"), textprops = {'fontsize': 12})centre = plt.Circle((0,0), 0.45, fc = "white", edgecolor = "black")plt.gcf().gca().add_artist(centre)# Histogramcountplt = plt.subplot(1, 2, 2)plt.title("Histogram", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "customer_type", data = df, order = order, edgecolor = "black")for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Customer Type", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df['customer_type'].value_counts()

? 结论：大多数人没有选择跟团旅游。

? 押金情况分析

plt.figure(figsize = (19, 12))order = sorted(df["deposit_type"].unique())plt.title("Bookings By Deposit Types", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "deposit_type", data = df, hue = "hotel", edgecolor = "black", palette = "bone", order = order)plt.xlabel("Deposit Type", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df["deposit_type"].value_counts()

? 结论：大部分客人没有交押金。

? 客人类型分析

labels = ['New Guest', 'Repeated Guest']colors = ["#00008B", "#C1CDCD"]order = df['is_repeated_guest'].value_counts().indexplt.figure(figsize = (19, 9))plt.suptitle('Bookings By Type Of Guest', fontweight = 'heavy', fontsize = '16',            fontfamily = 'sans-serif', color = "black")# Pie Chartplt.subplot(1, 2, 1)plt.title('Pie Chart', fontweight = 'bold', fontfamily = "sans-serif", color = 'black')plt.pie(df["is_repeated_guest"].value_counts(), pctdistance = 0.7, autopct = '%.2f%%', labels = labels,wedgeprops = dict(alpha = 0.8, edgecolor = "black"), textprops = {'fontsize': 12}, colors = colors)centre = plt.Circle((0,0), 0.45, fc = "white", edgecolor = "black")plt.gcf().gca().add_artist(centre)# Histogramcountplt = plt.subplot(1, 2, 2)plt.title("Histogram", fontweight = "bold", fontsize = 14, fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "is_repeated_guest", data = df, order = order, edgecolor = "black", palette = colors)for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Type Of Guest", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Total", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.xticks([0, 1], labels)plt.grid(axis = "y", alpha = 0.4)df['is_repeated_guest'].value_counts()

? 结论：几乎所有的客人都是新客人。

? 预订房间类型分析

plt.figure(figsize = (19, 12))order = sorted(df["reserved_room_type"].unique())plt.title("Bookings By Reserved Room Types", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "reserved_room_type", data = df, hue = "hotel", edgecolor = "black", palette = "bone", order = order)plt.xlabel("Reserved Room Type", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df["reserved_room_type"].value_counts()

? 结论：大多数客人预订了房间A，少数预订了房间D和E，其余的需求很少。

? 分配的房间类型分析

plt.figure(figsize = (19, 12))order = sorted(df["assigned_room_type"].unique())plt.title("Bookings By Assigned Room Types", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "assigned_room_type", data = df, hue = "hotel", edgecolor = "black", palette = "bone", order = order)plt.xlabel("Assigned Room Type", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df["assigned_room_type"].value_counts()

? 结论：大多数客人被分配到 A 室，少数被分配到 D 和 E 室，其余的很少。

? 预订状态分析

plt.figure(figsize = (19, 12))order = sorted(df["reservation_status"].unique())plt.title("Bookings By Reservation Status", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "reservation_status", data = df, hue = "hotel", edgecolor = "black", palette = "ocean", order = order)plt.xlabel("Reservation Status", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df["reservation_status"].value_counts()

? 结论：大多数客人登记入住并已经离开。

? 总住宿夜数分布

plt.figure(figsize = (19, 9))df2 = df.groupby("total_nights")["total_nights"].count()df2.sort_values(ascending = False)[: 10].plot(kind = 'bar')plt.title("Bookings By Total Nights Stayed By Guests", fontweight = "bold", fontsize = 14, fontfamily = "sans-serif", color = 'black')plt.xticks(rotation = 30)plt.xlabel("Number Of Nights", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)

? 结论：最受欢迎的酒店住宿时间是三晚。

? 酒店&总住宿夜数

plt.figure(figsize = (19, 12))order = df.total_nights.value_counts().iloc[:10].indexplt.title("Total Nights Stayed By Guests In Hotel", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "total_nights", data = df, hue = "hotel", edgecolor = "black", palette = "ocean", order = order)plt.xlabel("Total Nights", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)

? 结论：度假酒店最受欢迎的住宿时间是一晚、七晚、两晚、三晚和四晚。城市酒店最受欢迎的住宿时间是三晚、两晚、一晚和四晚。

? 热门国家分布

plt.figure(figsize = (19, 9))df2 = df.groupby("country")["country"].count()df2.sort_values(ascending = False)[: 20].plot(kind = 'bar')plt.title("Bookings By Top 20 Countries", fontweight = "bold", fontsize = 14, fontfamily = "sans-serif", color = 'black')plt.xticks(rotation = 30)plt.xlabel("Country", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df["country"].value_counts()

? 结论：在这份数据中，葡萄牙的预订量比其他任何国家都多。

? 预定下单时间

plt.figure(figsize = (16, 6))plt.title("Bookings By Lead Time", fontweight = "bold", fontsize = 14, fontfamily = 'sans-serif', color = 'black')sns.histplot(data = df, x = 'lead_time', hue = "hotel", kde = True, color = "#104E8B")plt.xlabel('Lead Time', fontweight = 'normal', fontsize = 11, fontfamily = 'sans-serif', color = "black")plt.ylabel('Number Of Bookings', fontweight = 'regular', fontsize = 11, fontfamily = "sans-serif", color = "black")df["lead_time"].describe().T

? 结论：大多数预订是在入住酒店前 100 天内完成的。

? 关联分析? 预订取消&酒店类型

plt.figure(figsize = (19, 12))plt.title("Number Of Bookings Cancelled By Guests", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "hotel", data = df, hue = "is_canceled", edgecolor = "black", palette = "bone")for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Hotel", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)

? 结论：度假村酒店的客人取消预订的频率低于城市酒店的客人。

? 预约取消&新老客

plt.figure(figsize = (19, 12))plt.title("Number Of Bookings Cancelled By Type Of Guests", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "is_canceled", data = df, hue = "is_repeated_guest", edgecolor = "black", palette = "bone")for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Cancellation", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.legend(['New Guest', 'Repeated Guest'])plt.grid(axis = "y", alpha = 0.4)

? 结论：老客取消预订的次数少于新客。

? 预约取消&细分市场

plt.figure(figsize = (19, 12))plt.title("Number Of Bookings Cancelled By Market Segments", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "market_segment", data = df, hue = "is_canceled", edgecolor = "black", palette = "bone")for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Market Segment", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)

? 结论：在线旅行社、线下旅行社/运营商和直销部分的取消率高于其他部分。

? 预订数量&年份

plt.figure(figsize = (19, 12))plt.title("Number Of Bookings Per Year", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "arrival_date_year", data = df, hue = "hotel", edgecolor = "black", palette = "cool")for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Year", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df["arrival_date_year"].value_counts()

? 结论：度假村和城市酒店在 2016 年的预订量均最高。与度假村酒店相比，城市酒店在 2017 年的预订量更高。两者在 2015 年的预订量几乎相同。

? 预订数量&月份

plt.figure(figsize = (16, 10))plt.title("Number Of Bookings Per Customer Type", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "customer_type", data = df, hue = "hotel", edgecolor = "black", palette = "pink")for rect in ax.patches:    ax.text(rect.get_x() + rect.get_width()/2, rect.get_height() + 4.25, rect.get_height(),    horizontalalignment="center", fontsize = 10, bbox = dict(facecolor = "none", edgecolor = "black",    linewidth = 0.25, boxstyle = "round"))plt.xlabel("Customer Type", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)

? 结论：十一月、十二月、一月和二月是预定最少的月份，7 月和 8 月是预订高峰月份。

? 预订数量&客户类型

months = ["January", "February", "March", "April", "May", "June", "July", "August", "September", "October","November", "December"]plt.figure(figsize = (19, 12))plt.title("Number Of Bookings Per Month", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')d = df.groupby("arrival_date_month")["arrival_date_month"].count()sns.barplot(x = d.index, y = d, order = months)plt.xticks(rotation = 30)plt.xlabel("Months")plt.ylabel("Number Of Bookings")df["arrival_date_month"].value_counts()

? 结论：临时和临时派对客人大多预订城市酒店，而跟团客人在度假村和城市酒店的预订数量几乎相同。

? 车位&预订

plt.figure(figsize = (19, 12))plt.title("Number Of Bookings Per Required Car Parking Space", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')ax = sns.countplot(x = "required_car_parking_spaces", data = df, hue = "hotel", edgecolor = "black", palette = "cool")plt.xlabel("Required Car Parking Space", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.ylabel("Number Of Bookings", fontweight = "bold", fontsize = 11, fontfamily = "sans-serif", color = "black")plt.grid(axis = "y", alpha = 0.4)df["required_car_parking_spaces"].value_counts()

? 结论：大多数客人不需要停车位，而少数客人需要停车位。

? 国家/地区&特殊要求数量

df2 = df.groupby("country")["total_of_special_requests"].mean().sort_values(ascending = False)[: 20]plt.figure(figsize = (18, 8))sns.barplot(x = df2.index, y = df2)plt.xticks(rotation = 30)plt.xlabel("Country")plt.ylabel("Average Number Of Special Requests")plt.title("Average Number Of Special Requests Made By Top 20 Countries ", fontweight = "bold", fontsize = 14,fontfamily = "sans-serif", color = 'black')

? 结论：在这些国家中，博茨瓦纳的特殊要求数量最多。

? 客户类型&特殊要求数量

df2 = df.groupby("customer_type")["total_of_special_requests"].mean().sort_values(ascending = False)[: 20]plt.figure(figsize = (18, 8))sns.barplot(x = df2.index, y = df2)plt.xticks(rotation = 30)plt.xlabel("Customer Type")plt.ylabel("Average Number Of Special Requests")plt.title("Average Number Of Special Requests By Customer Type", fontweight = "bold", fontsize = 14, fontfamily = "sans-serif", color = 'black')

? 结论：跟团客人的特殊要求数量最多，而临时派对客人的特殊要求数量最少。

? 月份&特殊要求数量

months = ["January", "February", "March", "April", "May", "June", "July", "August", "September", "October","November", "December"]df2 = df.groupby("arrival_date_month")["total_of_special_requests"].mean().sort_values(ascending = False)[: 20]plt.figure(figsize = (18, 8))sns.barplot(x = df2.index, y = df2, order = months)plt.xticks(rotation = 30)plt.xlabel("Months")plt.ylabel("Average Number Of Special Requests")plt.title("Average Number Of Special Requests By Guests Per Months ", fontweight = "bold", fontsize = 14, fontfamily = "sans-serif", color = 'black')

? 结论：客人在几个月内提出了几乎相似数量的特殊要求，但在 8 月、7 月和 12 月提出的特殊要求略多一些。

? 酒店类型&价格

# Histogramfig = plt.figure(figsize = (16, 10))df.drop(df[df["adr"] == 5400].index, axis = 0, inplace = True)plt.suptitle("Average Daily Rate Per Hotel", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')plot1 = fig.add_subplot(1, 2, 2)plt.title("Histogram Plot", fontweight = "bold", fontsize = 14, fontfamily = 'sans-serif', color = 'black')sns.histplot(data = df, x = 'adr', hue = "hotel", kde = True, color = "#104E8B")plt.xlabel('Average Daily Rate', fontweight = 'normal', fontsize = 11, fontfamily = 'sans-serif', color = "black")plt.ylabel('Count', fontweight = 'regular', fontsize = 11, fontfamily = "sans-serif", color = "black")# Box Plotplot2 = fig.add_subplot(1, 2, 1)plt.title("Box Plot", fontweight = "bold", fontsize = 14, fontfamily = 'sans-serif', color = 'black')sns.boxplot(data = df, x = "hotel", y = 'adr', color = "#104E8B")plt.ylabel('Average Daily Rate', fontweight = 'regular', fontsize = 11, fontfamily = 'sans-serif', color = "black")plt.show()df["adr"].describe()

? 结论：度假村酒店的平均每日价格比城市酒店更分散。

? 月份&费率

months = ["January", "February", "March", "April", "May", "June", "July", "August", "September", "October","November", "December"]df.drop(df[df["adr"] == 5400].index, axis = 0, inplace = True)d = df.groupby(["hotel", "arrival_date_month"])["adr"].mean().reset_index()d["arrival_date_month"] = pd.Categorical(d["arrival_date_month"], categories = months, ordered = True)d.sort_values("arrival_date_month", inplace = True)fig = plt.figure(figsize = (16, 10))plt.suptitle("Average Daily Rate Per Month", fontweight = "bold", fontsize = 14,         fontfamily = "sans-serif", color = 'black')sns.lineplot(data = d, y = 'adr', x = "arrival_date_month", hue = "hotel")plt.ylabel('Average Daily Rate', fontweight = 'normal', fontsize = 11, fontfamily = 'sans-serif', color = "black")plt.xlabel('Months', fontweight = 'regular', fontsize = 11, fontfamily = "sans-serif", color = "black")plt.xticks(rotation = 30)

? 结论：两类酒店的平均每日房价在年中均较高。与度假村酒店相比，城市酒店在年初和年末的每日房价较高。

? 相关性分析? 相关矩阵

我们计算一下相关矩阵，看看字段间的相关性如何

# 剔除一些不参与相关分析的字段df_sub = df.drop(['arrival_date_week_number', 'arrival_date_day_of_month', 'previous_cancellations','previous_bookings_not_canceled', 'booking_changes', 'reservation_status_date', 'agent', 'company', 'days_in_waiting_list', 'adults', 'babies', 'children'], axis = 1)# 相关矩阵corr_matrix = round(df_sub.corr(), 3)"Correlation Matrix: "corr_matrix

? 热力图

我们做一个热力图的绘制，以便更清晰看到字段间相关性。

plt.rcParams['figure.figsize'] =(12, 6)sns.heatmap(df_sub.corr(), annot=True, cmap='Reds', linewidths=5)plt.suptitle('Correlation Between Variables', fontweight='heavy', x=0.03, y=0.98, ha = "left", fontsize='18', fontfamily='sans-serif', color= "black")

参考资料

  • ? 数据科学工具库速查表 | Pandas 速查表：https://www.showmeai.tech/article-detail/101
  • ? 图解数据分析：从入门到精通系列教程：https://www.showmeai.tech/tutorials/33