APP Store评分案例

最新推荐文章于 2021-11-28 17:59:48 发布

Luara_lyy

最新推荐文章于 2021-11-28 17:59:48 发布

阅读量934

点赞数

分类专栏：商业数据分析师

本文链接：https://blog.csdn.net/weixin_40801364/article/details/100602493

版权

商业数据分析师专栏收录该内容

12 篇文章 1 订阅

订阅专栏

#调用基本包
import pandas as pd

#数据读取
app=pd.read_csv('w1_applestore.csv')

#数据的基本信息
app.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7197 entries, 0 to 7196
Data columns (total 11 columns):
Unnamed: 0          7197 non-null int64
id                  7197 non-null int64
track_name          7197 non-null object
size_bytes          7197 non-null int64
price               7197 non-null float64
rating_count_tot    7197 non-null int64
user_rating         7197 non-null float64
prime_genre         7197 non-null object
sup_devices         7197 non-null int64
ipadSc_urls         7197 non-null int64
lang                7197 non-null int64
dtypes: float64(2), int64(7), object(2)
memory usage: 618.6+ KB

app.head()

	Unnamed: 0	id	track_name	size_bytes	price	rating_count_tot	user_rating	prime_genre	sup_devices	ipadSc_urls	lang
0	0	281656475	PAC-MAN Premium	100788224	3.99	21292	4.0	Games	38	5	10
1	1	281796108	Evernote - stay organized	158578688	0.00	161065	4.0	Productivity	37	5	23
2	2	281940292	WeatherBug - Local Weather, Radar, Maps, Alerts	100524032	0.00	188583	3.5	Weather	37	5	3
3	3	282614216	eBay: Best App to Buy, Sell, Save! Online Shop...	128512000	0.00	262241	4.0	Shopping	37	5	9
4	4	282935706	Bible	92774400	0.00	985920	4.5	Reference	37	5	45

#发现了unname 0这个奇怪的变量，需要进行清理
app.drop('Unnamed: 0',axis=1,inplace=True)
#drop默认是对行，axis=1为列
#inplace表示直接替换掉原有数据
#同样可以用位置来举
#app.drop(app.columns[0],axis=1,inplace=True)

#对数据型变量的整体描述，只对数值型有效
app.describe()
#数据的集中度：平均数，众数，中位数
#离散度：方差，标准差，四分位数

	id	size_bytes	price	rating_count_tot	user_rating	sup_devices	ipadSc_urls	lang
count	7.197000e+03	7.197000e+03	7197.000000	7.197000e+03	7197.000000	7197.000000	7197.000000	7197.000000
mean	8.631310e+08	1.991345e+08	1.726218	1.289291e+04	3.526956	37.361817	3.707100	5.434903
std	2.712368e+08	3.592069e+08	5.833006	7.573941e+04	1.517948	3.737715	1.986005	7.919593
min	2.816565e+08	5.898240e+05	0.000000	0.000000e+00	0.000000	9.000000	0.000000	0.000000
25%	6.000937e+08	4.692275e+07	0.000000	2.800000e+01	3.500000	37.000000	3.000000	1.000000
50%	9.781482e+08	9.715302e+07	0.000000	3.000000e+02	4.000000	37.000000	5.000000	1.000000
75%	1.082310e+09	1.819249e+08	1.990000	2.793000e+03	4.500000	38.000000	5.000000	8.000000
max	1.188376e+09	4.025970e+09	299.990000	2.974676e+06	5.000000	47.000000	5.000000	75.000000

# 考虑将sizebytes变成mb，新增数据
app['size_mb'] = app['size_bytes'] / (1024 * 1024.0)

app.size_mb.describe()

count    7197.000000
mean      189.909414
std       342.566408
min         0.562500
25%        44.749023
50%        92.652344
75%       173.497070
max      3839.463867
Name: size_mb, dtype: float64

# 根据价格新增标签
app['paid'] = app['price'].apply(lambda x: 1 if x > 0 else 0)
#lambda阐述规则，X为price，为paid赋值，即当price＞0，paid为1，其他情况下，paid为0

app.paid.describe()

count    7197.000000
mean        0.436432
std         0.495977
min         0.000000
25%         0.000000
50%         0.000000
75%         1.000000
max         1.000000
Name: paid, dtype: float64

# 该部分小结,初步完成数据清洗
#清洗异常值（unamed)
#给分析造成难度的值(size-bytes)
#核心变量的创建（免费/收费)

#value_counts (price,prime_genre)
#value_Coutn只能对应series，不能对整个dataframe做操作
app.price.value_counts()

0.00      4056
0.99       728
2.99       683
1.99       621
4.99       394
3.99       277
6.99       166
9.99        81
5.99        52
7.99        33
14.99       21
19.99       13
8.99         9
24.99        8
13.99        6
11.99        6
29.99        6
12.99        5
15.99        4
59.99        3
17.99        3
22.99        2
23.99        2
20.99        2
27.99        2
16.99        2
49.99        2
39.99        2
74.99        1
18.99        1
34.99        1
99.99        1
299.99       1
47.99        1
21.99        1
249.99       1
Name: price, dtype: int64

#删除价格大于等于49.99的app
app=app[app['price']<=49.99]

#数据的快速分组
bins = [0,2,10,300]
labels = [  '<2', '<10','<300']
app['price_new']=pd.cut(app.price, bins, right=False, labels=labels)

app.groupby(['price_new'])['price'].describe()

	count	mean	std	min	25%	50%	75%	max
price_new
<2	5405.0	0.361981	0.675318	0.00	0.00	0.00	0.00	1.99
<10	1695.0	4.565811	1.864034	2.99	2.99	3.99	4.99	9.99
<300	90.0	20.256667	8.245939	11.99	14.99	17.99	23.99	49.99

# groupby的操作,不同类别app的价格分布
app.groupby(['prime_genre'])['price'].describe()

	count	mean	std	min	25%	50%	75%	max
prime_genre
Book	112.0	1.790536	3.342210	0.0	0.0	0.00	2.99	27.99
Business	56.0	4.136429	7.154403	0.0	0.0	2.99	4.99	49.99
Catalogs	10.0	0.799000	2.526660	0.0	0.0	0.00	0.00	7.99
Education	449.0	2.572004	2.669539	0.0	0.0	2.99	2.99	24.99
Entertainment	535.0	0.889701	1.454022	0.0	0.0	0.00	1.99	9.99
Finance	104.0	0.421154	1.108990	0.0	0.0	0.00	0.00	5.99
Food & Drink	63.0	1.552381	3.972119	0.0	0.0	0.00	1.49	27.99
Games	3862.0	1.432923	2.486609	0.0	0.0	0.00	1.99	29.99
Health & Fitness	180.0	1.916444	2.052378	0.0	0.0	1.99	2.99	9.99
Lifestyle	144.0	0.885417	1.478410	0.0	0.0	0.00	1.24	4.99
Medical	23.0	8.776087	10.788269	0.0	0.0	3.99	16.99	34.99
Music	138.0	4.835435	8.915667	0.0	0.0	0.99	4.99	49.99
Navigation	45.0	2.550000	4.487584	0.0	0.0	0.99	2.99	20.99
News	75.0	0.517733	1.127771	0.0	0.0	0.00	0.00	3.99
Photo & Video	349.0	1.473295	2.280703	0.0	0.0	0.99	1.99	22.99
Productivity	177.0	3.790113	4.965777	0.0	0.0	1.99	4.99	29.99
Reference	64.0	4.836875	8.285100	0.0	0.0	1.99	4.99	47.99
Shopping	122.0	0.016311	0.180166	0.0	0.0	0.00	0.00	1.99
Social Networking	167.0	0.339880	1.142210	0.0	0.0	0.00	0.00	9.99
Sports	114.0	0.953070	2.419084	0.0	0.0	0.00	0.99	19.99
Travel	81.0	1.120370	2.183772	0.0	0.0	0.00	0.99	9.99
Utilities	248.0	1.647621	2.628541	0.0	0.0	0.99	1.99	24.99
Weather	72.0	1.605417	1.831316	0.0	0.0	0.99	2.99	9.99

app.rating_count_tot.describe()

count    7.190000e+03
mean     1.290515e+04
std      7.577526e+04
min      0.000000e+00
25%      2.725000e+01
50%      3.005000e+02
75%      2.796750e+03
max      2.974676e+06
Name: rating_count_tot, dtype: float64

#对用户打分的分组
bins = [0,1000,5000,100000,5000000]
app['rating_new']=pd.cut(app.rating_count_tot, bins, right=False)

#用户打分和价格的关系
app.groupby(['rating_new'])['price'].describe()

---------------------------------------------------------------------------

NameError                                 Traceback (most recent call last)

<ipython-input-1-7db46c2c9387> in <module>()
      1 #用户打分和价格的关系
----> 2 app.groupby(['rating_new'])['price'].describe()


NameError: name 'app' is not defined

[外链图片转存失败(img-MmcB1nyr-1567848025818)(attachment:_auto_0)]

#可视化部分
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline

#app评分关系
plt.figure(figsize=(30,20))#调整大小
sns.relplot(x="prime_genre", y="user_rating",kind='line',
              data=app) #折线图

![在这里插入图片描述](https://img-blog.csdnimg.cn/20190907172142895.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl80MDgwMTM2NA==,size_16,color_FFFFFF,t_70)

app1=app[app['price']<=9.99]

#直方图，APP价格的分布
sns.distplot(app1['price'])

<matplotlib.axes._subplots.AxesSubplot at 0x197e7afd4e0>

[外链图片转存失败(img-Jkqtoe5D-1567848025819)(output_24_1.png)]

#业务问题2：收费app的价格分布是如何的？不同类别之间有关系吗？

#箱线图，不同类别APP价格
plt.figure(figsize=(10,8))#调整大小
sns.boxplot(x='price',y='prime_genre',data=app[app['paid']==1])

#业务解答：价格绝大部分都集中在9.99美元以内，个别类别（如医疗）等因专业性总体价格会高于其他类别

<matplotlib.axes._subplots.AxesSubplot at 0x197e7905320>

[外链图片转存失败(img-Rus1wMa9-1567848025819)(output_25_1.png)]

#只保留五个类别数据
top5= ['Games','Entertainment', 'Education', 'Photo & Video',
       'Utilities']
app5 = app[app.prime_genre.isin(top5)]

#箱线图，前五个类别的app价格

plt.figure(figsize=(10,8))#调整大小
sns.boxplot(x='price',y='prime_genre',data=app5[app['paid']==1])

C:\Anaconda3\lib\site-packages\ipykernel_launcher.py:4: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  after removing the cwd from sys.path.





<matplotlib.axes._subplots.AxesSubplot at 0x197e97f3518>

[外链图片转存失败(img-UfxBAMNM-1567848025819)(output_27_2.png)]

#散点图，价格和用户评分的分布
sns.scatterplot(x='price',y='user_rating',data=app)

<matplotlib.axes._subplots.AxesSubplot at 0x197e81ae710>

[外链图片转存失败(img-GpeoFmNH-1567848025819)(output_28_1.png)]

#只保留五个类别数据
top5= ['Games','Entertainment', 'Education', 'Photo & Video',
       'Utilities']
app5 = app[app.prime_genre.isin(top5)]

#柱状图，前5个类别app的用户评分均值
#同一类别，将免费和付费的评分进行对比
plt.figure(figsize=(10,8))
sns.barplot(x='prime_genre',y='user_rating',hue='paid',data=app5)

<matplotlib.axes._subplots.AxesSubplot at 0x197e7fece48>

[外链图片转存失败(img-GvNqLeCi-1567848025820)(output_30_1.png)]

#业务问题1 免费或收费APP集中在哪些类别
#第一步，将数据加总成每个类别有多少个app
#第二步，从高到低进行排列
#第三步，将数据进行可视化

#使用countplot--count是对数据加总，plot将数据进行可视化
#使用order对数据进行排序

plt.figure(figsize=(20,10))
sns.countplot(y='prime_genre',hue='paid',data=app,order=app['prime_genre'].value_counts().index)
plt.tick_params(labelsize=20)
#index是转列
#业务解答：都是高度集中在游戏类别

[外链图片转存失败(img-R7wv07IO-1567848025820)(output_31_0.png)]

#免费与收费的APP在不同评分区间的分布
bins=[0,0.1,2.5,4.5,5]
app['rating_level']=pd.cut(app.user_rating,bins,right=False)

app.groupby(['rating_level'])['user_rating'].describe()

	count	mean	std	min	25%	50%	75%	max
rating_level
[0.0, 0.1)	929.0	0.000000	0.000000	0.0	0.0	0.0	0.0	0.0
[0.1, 2.5)	206.0	1.650485	0.400213	1.0	1.5	2.0	2.0	2.0
[2.5, 4.5)	2903.0	3.646056	0.467987	2.5	3.5	4.0	4.0	4.0
[4.5, 5.0)	2660.0	4.500000	0.000000	4.5	4.5	4.5	4.5	4.5

sns.countplot(x='paid',hue='rating_level',data=app)

<matplotlib.axes._subplots.AxesSubplot at 0x197e9577f28>

[外链图片转存失败(img-AKjMqnn5-1567848025820)(output_34_1.png)]

#业务问题3：APP的大小和用户评分之间有关系吗？
q4=['user_rating','price','size_mb']

app[q4].corr()

	user_rating	price	size_mb
user_rating	1.000000	0.073237	0.066160
price	0.073237	1.000000	0.314386
size_mb	0.066160	0.314386	1.000000

#业务解答：大小价格都不和评分没有直接关系，但是价格和大小之间有正相关关系

sns.heatmap(app[q4].corr())
#热力图，展现变量之间两两之间关系的强弱

<matplotlib.axes._subplots.AxesSubplot at 0x197e95ce6d8>

[外链图片转存失败(img-gfcZseez-1567848025820)(output_38_1.png)]

Luara_lyy

关注

0
点赞
踩
6

收藏

觉得还不错? 一键收藏
0
评论
APP Store评分案例

#调用基本包import pandas as pd#数据读取app=pd.read_csv('w1_applestore.csv')#数据的基本信息app.info()<class 'pandas.core.frame.DataFrame'>RangeIndex: 7197 entries, 0 to 7196Data columns (total 11 colu...
复制链接

扫一扫

专栏目录