BikeSharing count predict

已于 2024-05-25 20:38:54 修改
阅读量756 收藏 17
点赞数 14
分类专栏： 机器学习共享单车数据处理 文章标签： scikit-learn
于 2024-05-25 20:38:11 首次发布
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/m0_73519831/article/details/139202969
版权

共享单车数据预测

哈喽大家，本次第二期呢是给大家分享一下最近做的关于共享单车的实验。下面我们开始吧。

1.导入数据

import pandas as pd
import numpy as np

data_Train=pd.read_csv(r'E:\Jupyter\ShareBike\data\train.csv')
data_Test=pd.read_csv(r'E:\Jupyter\ShareBike\data\test.csv')

data_Train.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 10886 entries, 0 to 10885
Data columns (total 12 columns):
datetime      10886 non-null object
season        10886 non-null int64
holiday       10886 non-null int64
workingday    10886 non-null int64
weather       10886 non-null int64
temp          10886 non-null float64
atemp         10886 non-null float64
humidity      10886 non-null int64
windspeed     10886 non-null float64
casual        10886 non-null int64
registered    10886 non-null int64
count         10886 non-null int64
dtypes: float64(3), int64(8), object(1)
memory usage: 1020.6+ KB

data_Test.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 6493 entries, 0 to 6492
Data columns (total 9 columns):
datetime      6493 non-null object
season        6493 non-null int64
holiday       6493 non-null int64
workingday    6493 non-null int64
weather       6493 non-null int64
temp          6493 non-null float64
atemp         6493 non-null float64
humidity      6493 non-null int64
windspeed     6493 non-null float64
dtypes: float64(3), int64(5), object(1)
memory usage: 456.6+ KB

可以看到数据完整并无缺失

2.特征工程

表中数据特征的实际意义如下：

  • datetime - hourly date + timestamp
  • season - 1 = spring, 2 = summer, 3 = fall, 4 = winter
  • holiday - whether the day is considered a holiday
  • workingday - whether the day is neither a weekend nor holiday
  • weather -
    • 1: Clear, Few clouds, Partly cloudy, Partly cloudy
    • 2: Mist + Cloudy, Mist + Broken clouds, Mist + Few clouds, Mist
    • 3: Light Snow, Light Rain + Thunderstorm + Scattered clouds, Light Rain + Scattered clouds
    • 4: Heavy Rain + Ice Pallets + Thunderstorm + Mist, Snow + Fog
  • temp - temperature in Celsius
  • atemp - “feels like” temperature in Celsius
  • humidity - relative humidity
  • windspeed - wind speed
  • casual - number of non-registered user rentals initiated
  • registered - number of registered user rentals initiated
  • count - number of total rentals

特征分析

显然，season/holiday/workingday/weather是范围型数据，而其当前的格式为int。
在后期进行数据可视化分析的时候，不够直观
因此尝试对这些数据进行修改：

import pylab
import calendar
import seaborn as sns
from scipy import stats
from datetime import datetime
import matplotlib.pyplot as plt
import warnings
warnings.filterwarnings('ignore',category=DeprecationWarning)

df=data_Train.append(data_Test)
df.reset_index(inplace=True)

C:\Users\lenovo\AppData\Roaming\Python\Python36\site-packages\pandas\core\frame.py:6211: FutureWarning: Sorting because non-concatenation axis is not aligned. A future version
of pandas will change to not sort by default.

To accept the future behavior, pass 'sort=False'.

To retain the current behavior and silence the warning, pass 'sort=True'.

  sort=sort)

df[['weather','season','holiday','workingday']].head()
weatherseasonholidayworkingday
01100
11100
21100
31100
41100
#将日期符号中的‘-’替换为‘/’,方便统一分析
df['datetime']=df['datetime'].map(lambda x:x.replace('-','/'))
#日期拆分成年月日
df['date']=df.datetime.apply(lambda x:x.split( )[0])
#拆分时间为小时
df['hour']=df.datetime.apply(lambda x:x.split( )[1].split(':')[0])
#准确定义每天为周几
df['weekday']=df.date.apply(lambda x:calendar.day_name[datetime.strptime(x,"%Y/%m/%d").weekday()])
df.weekday.value_counts()
#转化月份数据为直观的月份的数据
df['month']=df.date.apply(lambda x:calendar.month_name[datetime.strptime(x,"%Y/%m/%d").month])
#替换季节的数据为直观的季节
df['season']=df.season.map({1:'Spring',2:'Summer',3:'Fall',4:'Winter'})
#替换代表气象数据的数字为实际意义
df['weather']=df.weather.map({1: " Clear + Few clouds + Partly cloudy + Partly cloudy",\
                              2: " Mist + Cloudy, Mist + Broken clouds, Mist + Few clouds, Mist ", \
                              3: " Light Snow, Light Rain + Thunderstorm + Scattered clouds, Light Rain + Scattered clouds", \
                              4: " Heavy Rain + Ice Pallets + Thunderstorm + Mist, Snow + Fog " })

df=df.drop(['datetime'],axis=1)

df.drop('index',inplace=True,axis=1)

数据可视化

显然上一步得到的数据中无缺失值，
下面我们分析通过分析查看一般性规律并剔除异常值

#用sns绘制图像
fig, axes = plt.subplots(nrows=2,ncols=2)
fig.set_size_inches(12, 10)
sns.boxplot(data=df.loc[:10885],y="count",orient="v",ax=axes[0][0])
sns.boxplot(data=df.loc[:10885],y="count",x="season",orient="v",ax=axes[0][1])
sns.boxplot(data=df.loc[:10885],y="count",x="hour",orient="v",ax=axes[1][0])
sns.boxplot(data=df.loc[:10885],y="count",x="workingday",orient="v",ax=axes[1][1])

axes[0][0].set(ylabel='Count',title="Box Plot On Count")
axes[0][1].set(xlabel='Season', ylabel='Count',title="Box Plot On Count Across Season")
axes[1][0].set(xlabel='Hour Of The Day', ylabel='Count',title="Box Plot On Count Across Hour Of The Day")
axes[1][1].set(xlabel='Working Day', ylabel='Count',title="Box Plot On Count Across Working Day")

D:\Program\Anaconda\lib\site-packages\seaborn\categorical.py:462: FutureWarning: remove_na is deprecated and is a private function. Do not use.
  box_data = remove_na(group_data)





[Text(0,0.5,'Count'),
 Text(0.5,0,'Working Day'),
 Text(0.5,1,'Box Plot On Count Across Working Day')]

在这里插入图片描述

随后，通过matplot的形式绘图，进行对比，命令如下
可见matplot绘图相对比较麻烦，而且色彩设置比较繁琐

#整理数量与日期的关系，便于绘图
data=[]
Levels = df.season.unique()
for season in Levels:
    data.append(df.iloc[0:10885].loc[df.season==season,'count'])
    print(season)

Spring
Summer
Fall
Winter

#整理数量count与时间的关系，便于绘图
h_data=[]
hour=df.hour.loc[0:10885].unique()
for each in hour:
    h_data.append(df.iloc[0:10885].loc[df.hour==each,'count'])

#整理数量与工作日的关系，便于绘图
working_data=[]
working=df['workingday'].loc[0:10885].unique()
for each in working:
    working_data.append(df.iloc[0:10885].loc[df.workingday==each,'count'])

fig, axes = plt.subplots(2,2)
fig.set_size_inches(12, 10)

axes[0,0].boxplot(df['count'][:10885],labels=['count'],widths=0.5,patch_artist=True)
axes[0,1].boxplot(data,labels=['Spring','Summer','Fall','Winter'],patch_artist=True)
axes[1,0].boxplot(h_data,labels=hour,patch_artist=True)
axes[1,1].boxplot(working_data,labels=['workingday','weekend'],patch_artist=True,widths=0.5)

{'boxes': [<matplotlib.patches.PathPatch at 0x24eec686908>,
  <matplotlib.patches.PathPatch at 0x24eec65b518>],
 'caps': [<matplotlib.lines.Line2D at 0x24eec676e48>,
  <matplotlib.lines.Line2D at 0x24eec66ee10>,
  <matplotlib.lines.Line2D at 0x24eec0a4630>,
  <matplotlib.lines.Line2D at 0x24eec69e390>],
 'fliers': [<matplotlib.lines.Line2D at 0x24eec6865c0>,
  <matplotlib.lines.Line2D at 0x24eec69ec50>],
 'means': [],
 'medians': [<matplotlib.lines.Line2D at 0x24eec666eb8>,
  <matplotlib.lines.Line2D at 0x24eec69e7f0>],
 'whiskers': [<matplotlib.lines.Line2D at 0x24eec681908>,
  <matplotlib.lines.Line2D at 0x24eec681a58>,
  <matplotlib.lines.Line2D at 0x24eec656ba8>,
  <matplotlib.lines.Line2D at 0x24eec6569e8>]}

在这里插入图片描述

#除去偏差较大的数据
data_filtered=df.loc[:10885][np.abs(df['count']-df['count'].mean())<=(3*df['count'].std())]

D:\Program\Anaconda\lib\site-packages\ipykernel\__main__.py:2: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  from ipykernel import kernelapp as app

data_filtered
atempcasualcountholidayhumidityregisteredseasontempweatherwindspeedworkingdaydatehourweekdaymonth
014.3953.016.008113.0Spring9.84Clear + Few clouds + Partly cloudy + Partly c...0.000002011/1/10SaturdayJanuary
113.6358.040.008032.0Spring9.02Clear + Few clouds + Partly cloudy + Partly c...0.000002011/1/11SaturdayJanuary
213.6355.032.008027.0Spring9.02Clear + Few clouds + Partly cloudy + Partly c...0.000002011/1/12SaturdayJanuary
314.3953.013.007510.0Spring9.84Clear + Few clouds + Partly cloudy + Partly c...0.000002011/1/13SaturdayJanuary
414.3950.01.00751.0Spring9.84Clear + Few clouds + Partly cloudy + Partly c...0.000002011/1/14SaturdayJanuary
512.8800.01.00751.0Spring9.84Mist + Cloudy, Mist + Broken clouds, Mist + F...6.003202011/1/15SaturdayJanuary
613.6352.02.00800.0Spring9.02Clear + Few clouds + Partly cloudy + Partly c...0.000002011/1/16SaturdayJanuary
712.8801.03.00862.0Spring8.20Clear + Few clouds + Partly cloudy + Partly c...0.000002011/1/17SaturdayJanuary
814.3951.08.00757.0Spring9.84Clear + Few clouds + Partly cloudy + Partly c...0.000002011/1/18SaturdayJanuary
917.4258.014.00766.0Spring13.12Clear + Few clouds + Partly cloudy + Partly c...0.000002011/1/19SaturdayJanuary
1019.69512.036.007624.0Spring15.58Clear + Few clouds + Partly cloudy + Partly c...16.997902011/1/110SaturdayJanuary
1116.66526.056.008130.0Spring14.76Clear + Few clouds + Partly cloudy + Partly c...19.001202011/1/111SaturdayJanuary
1221.21029.084.007755.0Spring17.22Clear + Few clouds + Partly cloudy + Partly c...19.001202011/1/112SaturdayJanuary
1322.72547.094.007247.0Spring18.86Mist + Cloudy, Mist + Broken clouds, Mist + F...19.999502011/1/113SaturdayJanuary
1422.72535.0106.007271.0Spring18.86Mist + Cloudy, Mist + Broken clouds, Mist + F...19.001202011/1/114SaturdayJanuary
1521.97040.0110.007770.0Spring18.04Mist + Cloudy, Mist + Broken clouds, Mist + F...19.999502011/1/115SaturdayJanuary
1621.21041.093.008252.0Spring17.22Mist + Cloudy, Mist + Broken clouds, Mist + F...19.999502011/1/116SaturdayJanuary
1721.97015.067.008252.0Spring18.04Mist + Cloudy, Mist + Broken clouds, Mist + F...19.001202011/1/117SaturdayJanuary
1821.2109.035.008826.0Spring17.22Light Snow, Light Rain + Thunderstorm + Scatt...16.997902011/1/118SaturdayJanuary
1921.2106.037.008831.0Spring17.22Light Snow, Light Rain + Thunderstorm + Scatt...16.997902011/1/119SaturdayJanuary
2020.45511.036.008725.0Spring16.40Mist + Cloudy, Mist + Broken clouds, Mist + F...16.997902011/1/120SaturdayJanuary
2120.4553.034.008731.0Spring16.40Mist + Cloudy, Mist + Broken clouds, Mist + F...12.998002011/1/121SaturdayJanuary
2220.45511.028.009417.0Spring16.40Mist + Cloudy, Mist + Broken clouds, Mist + F...15.001302011/1/122SaturdayJanuary
2322.72515.039.008824.0Spring18.86Mist + Cloudy, Mist + Broken clouds, Mist + F...19.999502011/1/123SaturdayJanuary
2422.7254.017.008813.0Spring18.86Mist + Cloudy, Mist + Broken clouds, Mist + F...19.999502011/1/20SundayJanuary
2521.9701.017.009416.0Spring18.04Mist + Cloudy, Mist + Broken clouds, Mist + F...16.997902011/1/21SundayJanuary
2621.2101.09.001008.0Spring17.22Mist + Cloudy, Mist + Broken clouds, Mist + F...19.001202011/1/22SundayJanuary
2722.7252.06.00944.0Spring18.86Mist + Cloudy, Mist + Broken clouds, Mist + F...12.998002011/1/23SundayJanuary
2822.7252.03.00941.0Spring18.86Mist + Cloudy, Mist + Broken clouds, Mist + F...12.998002011/1/24SundayJanuary
2921.2100.02.00772.0Spring17.22Light Snow, Light Rain + Thunderstorm + Scatt...19.999502011/1/26SundayJanuary
................................................
1085619.69513.0525.0046512.0Winter15.58Clear + Few clouds + Partly cloudy + Partly c...22.002812012/12/1818TuesdayDecember
1085719.69519.0353.0046334.0Winter15.58Clear + Few clouds + Partly cloudy + Partly c...26.002712012/12/1819TuesdayDecember
1085816.6654.0268.0050264.0Winter14.76Clear + Few clouds + Partly cloudy + Partly c...16.997912012/12/1820TuesdayDecember
1085917.4259.0168.0050159.0Winter14.76Clear + Few clouds + Partly cloudy + Partly c...15.001312012/12/1821TuesdayDecember
1086016.6655.0132.0049127.0Winter13.94Clear + Few clouds + Partly cloudy + Partly c...0.000012012/12/1822TuesdayDecember
1086117.4251.081.004980.0Winter13.94Clear + Few clouds + Partly cloudy + Partly c...6.003212012/12/1823TuesdayDecember
1086215.9106.041.006135.0Winter12.30Clear + Few clouds + Partly cloudy + Partly c...0.000012012/12/190WednesdayDecember
1086315.9101.015.006514.0Winter12.30Clear + Few clouds + Partly cloudy + Partly c...6.003212012/12/191WednesdayDecember
1086415.1501.03.00652.0Winter11.48Clear + Few clouds + Partly cloudy + Partly c...6.003212012/12/192WednesdayDecember
1086513.6350.05.00755.0Winter10.66Clear + Few clouds + Partly cloudy + Partly c...8.998112012/12/193WednesdayDecember
1086612.1201.07.00756.0Winter9.84Clear + Few clouds + Partly cloudy + Partly c...8.998112012/12/194WednesdayDecember
1086714.3952.031.007529.0Winter10.66Clear + Few clouds + Partly cloudy + Partly c...6.003212012/12/195WednesdayDecember
1086812.8803.0112.0075109.0Winter9.84Clear + Few clouds + Partly cloudy + Partly c...6.003212012/12/196WednesdayDecember
1086913.6353.0363.0075360.0Winter10.66Clear + Few clouds + Partly cloudy + Partly c...8.998112012/12/197WednesdayDecember
1087012.88013.0678.0087665.0Winter9.84Clear + Few clouds + Partly cloudy + Partly c...7.001512012/12/198WednesdayDecember
1087114.3958.0317.0075309.0Winter11.48Clear + Few clouds + Partly cloudy + Partly c...7.001512012/12/199WednesdayDecember
1087216.66517.0164.0070147.0Winter13.12Clear + Few clouds + Partly cloudy + Partly c...7.001512012/12/1910WednesdayDecember
1087320.45531.0200.0054169.0Winter16.40Clear + Few clouds + Partly cloudy + Partly c...15.001312012/12/1911WednesdayDecember
1087420.45533.0236.0054203.0Winter16.40Clear + Few clouds + Partly cloudy + Partly c...19.001212012/12/1912WednesdayDecember
1087521.21030.0213.0050183.0Winter17.22Clear + Few clouds + Partly cloudy + Partly c...12.998012012/12/1913WednesdayDecember
1087621.21033.0218.0050185.0Winter17.22Clear + Few clouds + Partly cloudy + Partly c...12.998012012/12/1914WednesdayDecember
1087721.21028.0237.0050209.0Winter17.22Clear + Few clouds + Partly cloudy + Partly c...19.001212012/12/1915WednesdayDecember
1087821.21037.0334.0050297.0Winter17.22Clear + Few clouds + Partly cloudy + Partly c...23.999412012/12/1916WednesdayDecember
1087920.45526.0562.0050536.0Winter16.40Clear + Few clouds + Partly cloudy + Partly c...26.002712012/12/1917WednesdayDecember
1088019.69523.0569.0050546.0Winter15.58Clear + Few clouds + Partly cloudy + Partly c...23.999412012/12/1918WednesdayDecember
1088119.6957.0336.0050329.0Winter15.58Clear + Few clouds + Partly cloudy + Partly c...26.002712012/12/1919WednesdayDecember
1088217.42510.0241.0057231.0Winter14.76Clear + Few clouds + Partly cloudy + Partly c...15.001312012/12/1920WednesdayDecember
1088315.9104.0168.0061164.0Winter13.94Clear + Few clouds + Partly cloudy + Partly c...15.001312012/12/1921WednesdayDecember
1088417.42512.0129.0061117.0Winter13.94Clear + Few clouds + Partly cloudy + Partly c...6.003212012/12/1922WednesdayDecember
1088516.6654.088.006684.0Winter13.12Clear + Few clouds + Partly cloudy + Partly c...8.998112012/12/1923WednesdayDecember

10739 rows × 15 columns

#查看处理后的数据量
data_filtered.shape

(10739, 15)

3.数据相关性分析

上面，我们通过对日期等时间数据进行了处理，划分了月，日，小时，工作日等特征

随后，分析一下temp，atemp，humidity，windspeed与count，casual，registered之间的相关性

cor=df.loc[:10885][['temp','atemp','humidity','windspeed','casual','registered','count']].corr()

mask=np.array(cor)
mask

array([[ 1.        ,  0.98494811, -0.06494877, -0.01785201,  0.46709706,
         0.31857128,  0.39445364],
       [ 0.98494811,  1.        , -0.04353571, -0.057473  ,  0.46206654,
         0.31463539,  0.38978444],
       [-0.06494877, -0.04353571,  1.        , -0.31860699, -0.3481869 ,
        -0.26545787, -0.31737148],
       [-0.01785201, -0.057473  , -0.31860699,  1.        ,  0.09227619,
         0.09105166,  0.10136947],
       [ 0.46709706,  0.46206654, -0.3481869 ,  0.09227619,  1.        ,
         0.49724969,  0.69041357],
       [ 0.31857128,  0.31463539, -0.26545787,  0.09105166,  0.49724969,
         1.        ,  0.97094811],
       [ 0.39445364,  0.38978444, -0.31737148,  0.10136947,  0.69041357,
         0.97094811,  1.        ]])

mask[np.tril_indices_from(mask)]=0
mask

array([[ 0.        ,  0.98494811, -0.06494877, -0.01785201,  0.46709706,
         0.31857128,  0.39445364],
       [ 0.        ,  0.        , -0.04353571, -0.057473  ,  0.46206654,
         0.31463539,  0.38978444],
       [ 0.        ,  0.        ,  0.        , -0.31860699, -0.3481869 ,
        -0.26545787, -0.31737148],
       [ 0.        ,  0.        ,  0.        ,  0.        ,  0.09227619,
         0.09105166,  0.10136947],
       [ 0.        ,  0.        ,  0.        ,  0.        ,  0.        ,
         0.49724969,  0.69041357],
       [ 0.        ,  0.        ,  0.        ,  0.        ,  0.        ,
         0.        ,  0.97094811],
       [ 0.        ,  0.        ,  0.        ,  0.        ,  0.        ,
         0.        ,  0.        ]])

fig=plt.figure(figsize=(20,10))
sns.heatmap(cor,mask=mask,vmax=1,square=True,annot=True)

<matplotlib.axes._subplots.AxesSubplot at 0x24eec74d240>

在这里插入图片描述

fig,ax=plt.subplots(1,4,figsize=(20,8))
sns.regplot(x='temp',y='count',data=df.loc[:10885],ax=ax[0])
sns.regplot(x='atemp',y='count',data=df.loc[:10885],ax=ax[1])
sns.regplot(x='windspeed',y='count',data=df.loc[:10885],ax=ax[2])
sns.regplot(x='humidity',y='count',data=df.loc[:10885],ax=ax[3])

<matplotlib.axes._subplots.AxesSubplot at 0x24eec8c6128>

在这里插入图片描述

displot说明
heatmap说明

气温和体感温度的趋势基本一致

4.可视化数据

对count（结果）的数据分布分析
fig,ax=plt.subplots(2,2,figsize=(12,10))
sns.distplot(df.loc[:10885]['count'],ax=ax[0][0])
stats.probplot(df.loc[:10885]['count'],dist='norm',fit=True,plot=ax[0][1])
sns.distplot(np.log(df.loc[:10885]['count']),ax=ax[1][0])
stats.probplot(np.log(df.loc[:10885]['count']),dist='norm',fit=True,plot=ax[1][1])

((array([-3.83154229, -3.60754977, -3.48462983, ...,  3.48462983,
          3.60754977,  3.83154229]),
  array([ 0.        ,  0.        ,  0.        , ...,  6.87523209,
          6.87729607,  6.88448665])),
 (1.4251967988522145, 4.5525610951779747, 0.95684956593134451))

在这里插入图片描述

可以看到原有数据方差较大，这不利于最终的模型回归
为此，将数据进行对数化减小数据方差，
保证了数据近似于正态分布，不像之前的结果那样存在明显的向左偏差

可视化月份，季节，小时，星期、用户类型与数量的关系
#首先将时间重新转化成整型数据，方便后面对时间进行排序
df.hour=df.hour.astype(int)

fig,ax=plt.subplots(4,1,figsize=(12,20))

month = ["January","February","March","April","May","June","July","August","September","October","November","December"]
week = ["Sunday","Monday","Tuesday","Wednesday","Thursday","Friday","Saturday"]

monthAggregated=pd.DataFrame(df.loc[:10885].groupby('month')['count'].mean()).reset_index()
monthSorted=monthAggregated.sort_values(by='count',ascending=False)
sns.barplot(data=monthSorted,x='month',y='count',ax=ax[0],order=month)
ax[0].set(xlabel='Month',ylabel='Average Count',title='Average',label='big')

hourAggregated=pd.DataFrame(df.loc[:10885].groupby(['hour','season'],sort=True)['count'].mean()).reset_index().sort_values(by='hour')
sns.pointplot(x=hourAggregated['hour'],y=hourAggregated['count'],hue=hourAggregated["season"],data=hourAggregated, join=True,ax=ax[1])
ax[1].set(xlabel='Hour of the day',ylabel='Users Count',title="Average Users Count By Hour Of The Day Across Season",label='big')

hourAggregated=pd.DataFrame(df.loc[:10885].groupby(["hour","weekday"],sort=True)["count"].mean()).reset_index().sort_values(by='hour')
sns.pointplot(x=hourAggregated['hour'],y=hourAggregated['count'],
              hue=hourAggregated['weekday'],hue_order=week,
              data=hourAggregated,join=True,ax=ax[2])
ax[2].set(xlabel='Hour Of The Day', ylabel='Users Count',title="Average Users Count By Hour Of The Day Across Weekdays",label='big')

hourTransformed=pd.melt(df.loc[:10885][['hour','casual','registered']],id_vars=['hour'],value_vars=['casual','registered'])
hourAggregated=pd.DataFrame(hourTransformed.groupby(['hour','variable'],sort=True)['value'].mean()).reset_index().sort_values(by='hour')
sns.pointplot(x=hourAggregated['hour'],y=hourAggregated['value'],hue=hourAggregated['variable'],hue_order=['casual','registered'],data=hourAggregated,join=True)
ax[3].set(xlabel='Hour Of The Day', ylabel='Users Count',title="Average Users Count By Hour Of The Day Across User type",label='big')

D:\Program\Anaconda\lib\site-packages\seaborn\categorical.py:1460: FutureWarning: remove_na is deprecated and is a private function. Do not use.
  stat_data = remove_na(group_data)
D:\Program\Anaconda\lib\site-packages\seaborn\categorical.py:1508: FutureWarning: remove_na is deprecated and is a private function. Do not use.
  stat_data = remove_na(group_data[hue_mask])





[Text(0,0.5,'Users Count'),
 Text(0.5,0,'Hour Of The Day'),
 Text(0.5,1,'Average Users Count By Hour Of The Day Across User type'),
 None]

在这里插入图片描述

fig=plt.subplots(figsize=(5,5))
casual=df.loc[:10885,'casual'].sum()
registered=df.loc[:10885,'registered'].sum()
plt.pie([casual,registered],labels=['casual','registered'])
#plt.grid(True)
#plt.xlabel('user type')
#plt.ylabel('user count')
plt.title('user count compare between casual and registered users')

Text(0.5,1,'user count compare between casual and registered users')

在这里插入图片描述

显然，基于上述结果，我们得到如下规律：
1.用户用车数量春秋天明天高于冬夏，冬天的用车数量最少；
2.一天内，用车存在早晚两个高峰期，这与上下班的时间重合；
3.有别于工作日的用车规律，周末用车数量在早晨8点后开始上升，并在未来的10个小时内保持一个较高的数量；
4.显然注册用户的用车规律与工作日的用车曲线规律一致，这反映了注册用户普遍将共享单车作为了一种出行工具；
5.非注册用户则可能是出于娱乐等其他方面的需求。
6.注册用户数量明显高于非注册用户，约为4倍，这说明将单车用于工作的人比娱乐的人数要多；

5.修正windspeed数据

通过对数据进行观察发现，尽管数据完整，但在风速数据中，为0的数据超过1300个
显然，这部分数据信息有误，因此有必要根据已有的信息进行分析

windspeed=df.loc[:10885].windspeed
plt.bar(windspeed.unique(),windspeed.value_counts(),width=1,color="#87CEFA")
plt.xlabel('windspeed')
plt.ylabel('value_counts')
plt.grid(True)

在这里插入图片描述

显然，风速的结果与season，weather，humidity，month，temp，year，atemp有关。
而这其中的season，weather，month，year已经在之前处理过，
为进行预测，有必要将数据重新转化为数值型数据

df.season=df.season.map({'Spring':1,'Summer':2,'Fall':3,'Winter':4})

#转化数值类型便于进行模型回归
df.weather=df.weather.map({' Clear + Few clouds + Partly cloudy + Partly cloudy':1,
                ' Mist + Cloudy, Mist + Broken clouds, Mist + Few clouds, Mist ':2,
                ' Light Snow, Light Rain + Thunderstorm + Scattered clouds, Light Rain + Scattered clouds':3,
                ' Heavy Rain + Ice Pallets + Thunderstorm + Mist, Snow + Fog ':4})

df.month=df.month.map({'April': 4,'August': 8,'December': 12,'February': 2,
                       'January': 1,'July': 7,'June': 6,'March': 3,
                       'May': 5,'November': 11,'October': 10,'September': 9})

df['year']=df.date.apply(lambda x:x.split('/')[0])

df.weekday=df.weekday.map({'Saturday':6, 'Sunday':7, 'Monday':1, 'Tuesday':2, 'Wednesday':3, 'Thursday':4,'Friday':5})

df.head(1)
atempcasualcountholidayhumidityregisteredseasontempweatherwindspeedworkingdaydatehourweekdaymonthyear
014.3953.016.008113.019.8410.002011/1/10612011
#导入集成学习中的随机深林算法构建模型进行拟合
from sklearn.ensemble import RandomForestRegressor
datawind0=df[df['windspeed']==0]
datawindn0=df[df['windspeed']!=0]
model=RandomForestRegressor()

#模型拟合
windfactors=["season","weather","humidity","month","temp","year","atemp"]
model.fit(datawindn0[windfactors],datawindn0['windspeed'])

RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=None,
           max_features='auto', max_leaf_nodes=None,
           min_impurity_decrease=0.0, min_impurity_split=None,
           min_samples_leaf=1, min_samples_split=2,
           min_weight_fraction_leaf=0.0, n_estimators=10, n_jobs=1,
           oob_score=False, random_state=None, verbose=0, warm_start=False)

#风速结果预测
windvalue=model.predict(datawind0[windfactors])
datawind0['windspeed']=windvalue
data=datawindn0.append(datawind0)

D:\Program\Anaconda\lib\site-packages\ipykernel\__main__.py:3: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy
  app.launch_new_instance()

data.reset_index(inplace=True)
data.drop('index',inplace=True,axis=1)

显然，进行风速回归后的结果更加准确

6.结果预测

在上面对数据进行处理后，将数据拆分成训练集和测试集

data.head(1)
atempcasualcountholidayhumidityregisteredseasontempweatherwindspeedworkingdaydatehourweekdaymonthyear
012.880.01.00751.019.8426.003202011/1/15612011
dataTrain = data[pd.notnull(data['count'])].sort_values(by=['date','hour'])
dataTest = data[~pd.notnull(data['count'])].sort_values(by=['date','hour']) #~符号代表的是补集的意思,这里意思是‘count’为空数据集
datetimecol = data_Test['datetime'].replace('/','-')
yLabels = dataTrain["count"]
yLablesRegistered = dataTrain["registered"]
yLablesCasual = dataTrain["casual"]

dataTrain=dataTrain.drop(['casual','registered','count','date'],axis=1)
dataTest=dataTest.drop(['casual','registered','count','date'],axis=1)  

#本次竞赛官方提供的误差计算函数
def rmsle(y,y_,convertExp=True):
    if convertExp:
        y=np.exp(y)
        y_=np.exp(y_)
    log1=np.nan_to_num(np.array([np.log(v+1) for v in y]))
    log2=np.nan_to_num(np.array([np.log(v+1) for v in y_]))
    calc=(log1-log2)**2
    return np.sqrt(np.mean(calc))

线性回归模型分析

from sklearn.linear_model import LinearRegression,Ridge,Lasso
from sklearn.model_selection import GridSearchCV
from sklearn import metrics
import warnings
pd.options.mode.chained_assignment=None
warnings.filterwarnings('ignore',category=DeprecationWarning)

#初始化逻辑回归模型
model1=LinearRegression()

#训练模型
yLabelslog=np.log1p(yLabels)
model1.fit(dataTrain,yLabelslog)

#将当前的已有数据的预测值与实际值对比，计算误差。
#最终模拟结果也是这样评分的
preds=model1.predict(dataTrain)
print('RMSLE 值分析',rmsle(preds,yLabelslog))

RMSLE 值分析 0.977959172158

正则化模型-岭回归（Ridge）

#构建模型，ridge岭回归，GridsearchCV交叉验证获得不同alpha下的最优的拟合结果，ridge是二阶范数
model_r=Ridge()
ridge_params={'max_iter':[3000],'alpha':[0.1, 1, 2, 3, 4, 10, 30,100,200,300,400,800,900,1000]}
rmsle_scorer=metrics.make_scorer(rmsle,greater_is_better=False)
model_g=GridSearchCV(model_r,
                         ridge_params,
                         scoring=rmsle_scorer,
                         cv=5)

#训练模型
yLabelslog=np.log1p(yLabels)
model_g.fit(dataTrain,yLabelslog)

#查看模型的拟合效果
preds=model_g.predict(dataTrain)
print('RMSLE 值分析',rmsle(preds,yLabelslog))

RMSLE 值分析 0.978326250198

fig,ax=plt.subplots(figsize=(6,2.5))
df1=pd.DataFrame(model_g.grid_scores_)
alpha=df1.parameters.apply(lambda x:x['alpha'])
score=df1.mean_validation_score.apply(lambda x:-x)
sns.pointplot(alpha,score,ax=ax)

D:\Program\Anaconda\lib\site-packages\seaborn\categorical.py:1460: FutureWarning: remove_na is deprecated and is a private function. Do not use.
  stat_data = remove_na(group_data)





<matplotlib.axes._subplots.AxesSubplot at 0x24eee3a57f0>

在这里插入图片描述

正则化模型-Lasso

#构建模型，Lasso是一阶范数
model_l=Lasso()

alpha=1/np.array([0.1, 1, 2, 3, 4, 10, 30,100,200,300,400,800,900,1000])
lasso_params={'max_iter':[3000],'alpha':alpha}

model_gl=GridSearchCV(model_l,
                      lasso_params,
                     scoring=rmsle_scorer,
                     cv=5)

#训练模型
yLabelslog=np.log1p(yLabels)
model_gl.fit(dataTrain,yLabelslog)

#查看模型的拟合效果
preds=model_gl.predict(dataTrain)
print('RMSLE 值分析',rmsle(preds,yLabelslog))

RMSLE 值分析 0.978214656155

fig,ax=plt.subplots(figsize=(12,5))
df2=pd.DataFrame(model_gl.grid_scores_)
alpha=df2.parameters.apply(lambda x:x['alpha'])
score=df2.mean_validation_score.apply(lambda x:-x)
sns.pointplot(alpha,score,ax=ax)

D:\Program\Anaconda\lib\site-packages\seaborn\categorical.py:1460: FutureWarning: remove_na is deprecated and is a private function. Do not use.
  stat_data = remove_na(group_data)





<matplotlib.axes._subplots.AxesSubplot at 0x24eecf282b0>

在这里插入图片描述

集成化模型-RF

#为了直观说明，再次导入随机森林函数,构建模型
from sklearn.ensemble import RandomForestRegressor
model_RF=RandomForestRegressor(n_estimators=100)

#训练模型
yLabelslog=np.log1p(yLabels)
model_RF.fit(dataTrain,yLabelslog)

#查看模型的拟合效果
preds=model_RF.predict(dataTrain)
print('RMSLE 值分析',rmsle(preds,yLabelslog))

RMSLE 值分析 0.103272050217

集成化模型-梯度上升（Gradient Boost）

#导入模型，构建模型
from sklearn.ensemble import GradientBoostingRegressor
model_GB=GradientBoostingRegressor(n_estimators=4000,alpha=0.01)

#训练模型
yLabelslog=np.log1p(yLabels)
model_GB.fit(dataTrain,yLabelslog)

#查看模型的拟合效果
preds=model_GB.predict(dataTrain)
print('RMSLE 值分析',rmsle(preds,yLabelslog))

RMSLE 值分析 0.189423623632

显然，通过集成算法后的结果明显优于单模型计算的结果，
且从RMSLE的结果来看，随机森林（RF）的预测结果明显优于GB算法的结果

#获得预测结果
predsTest_RF=model_RF.predict(dataTest)
predsTest_GB=model_GB.predict(dataTest)

fig,ax=plt.subplots(3,1,figsize=(8,20))
sns.distplot(yLabels,ax=ax[0],label='真实值').legend()
sns.distplot(np.exp(predsTest_RF),ax=ax[1],label='预测值_RF').legend()
sns.distplot(np.exp(predsTest_GB),ax=ax[2],label='预测值_GB').legend()

<matplotlib.legend.Legend at 0x24eef180b38>

在这里插入图片描述

#生成数据格式
submission=pd.DataFrame({
                        'datetime':datetimecol,
                        'count':[max(0,x) for x in np.exp(predsTest_RF)]})

#导出数据
submission.to_csv(r'E:\Jupyter\ShareBike\data\Submission_RF.csv',index=False)

数据提交到kaggle，最终得分0.436分。
好了，本期的分享就到此结束，再见！

碎碎念念，岁岁年年
关注
  • 14
    点赞
  • 踩
  • 17
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
专栏目录
Bike Sharing 案例
zxymvp的博客
03-05 3628
通过对华盛顿的部分自行车租赁情况数据进行分析，对测试集的自行车租赁数目进行预测。 导入数据 train = read.csv("train.csv") test = read.csv("test.csv") str(train) str(test) 分别对两组数据进行概览，可以发现数据集存在以下特征： $ datetime : 日期 $ season : 季节，1—...
Bike sharing demand prediction 【kaggle competition】
anzhi的博客
06-28 1559
Bike sharing demand prediction
【数据分析实战】kaggle项目：bike sharing demand
SpriteNym的博客
08-22 1142
前10%，不戳！
kaggle入门-Bike Sharing Demand自行车需求预测
春华秋实
10-18 6813
接触机器学习断断续续有一年了，一直没有真正做点什么事，今天终于开始想刷刷kaggle的问题了，慢慢熟悉和理解机器学习以及深度学习。 今天第一题是一个比较基础的Bike Sharing Demand题，根据日期时间、天气、温度等特征，预测自行车的租借量。训练与测试数据集大概长这样： // train datetime,season,holiday,workingday,weather,temp,
Bike-sharing-demand
weixin_35980266的博客
08-15 1588
kaggle的竞赛题目 ：https://www.kaggle.com/c/bike-sharing-demand 引用文章 ： https://www.kaggle.com/c/bike-sharing-demand 1、问题的提出： Bike sharing systems are a means of renting bicycles where the process o
Kaggle - Bike Sharing Prediction
令狐公子的博客
02-13 568
import pylab import calendar import numpy as np import pandas as pd import seaborn as sn from scipy import stats import missingno as msno from datetime import datetime import matplotlib.pyplot as plt...
kaggle——Bike Sharing Demand
Do not mess it up
08-15 1133
个人感觉最主要是通过visualization来获取一些信息， 比如24小时变化的曲线 工作日24h变话的曲线 周末24h变换的曲线 从中就可以挖掘出一些有用的信息 挖掘出来最简单的应用就是根据时间来预测，用当前时间段的所有的平均值代替就行了， 但是这样做出来的结果是比较差的，还不如直接用random forest来的LB效果好 也许应该对不同的时间进
Bike Sharing Demand （Kaggle）
勿忘初衷
03-09 3715
Forecast use of a city bikeshare system Bike sharing systems are a means of renting bicycles where the process of obtaining membership, rental, and bike return is automated via a network of kiosk l
[索引]引用Balancing bike sharing systems with constraint programming的文章
panbaoran913的博客
12-19 1003
异构车队和多站点静态自行车重新平衡问题（HFMDSBRP-SD）是静态自行车重新均衡问题的扩展，考虑到异构车队多站点，允许拆分负载。它包括找到一组成本最低的重新定位车辆路线，并确定每个站点的取货或交货数量，以满足每个站点的需求。我们开发了一种分支和价格切割（BPC），其中引入了禁忌搜索列生成器和启发式标签设置算法，并将增强的初等不等式推广到该问题中，以提高全局收敛速度。计算结果证明了BPC算法的有效性。在最大大小为30的360个实例中，有298个实例能够在两小时的时间限制内实现最优。
sklearn的predict_proba使用说明
09-16
这与 `predict` 方法不同，`predict` 返回的是每个样本最有可能的类别标签，而 `predict_proba` 返回的是每个类别的概率分布。例如，在二分类问题中，`predict_proba` 返回的是一维数组，包含两个概率值，分别代表...
bp.zip_bp predict
07-14
在本压缩包“bp.zip_bp predict”中，包含了一个MATLAB实现的BP神经网络源代码，用于数据预测任务。 MATLAB是一种强大的数学计算环境，常用于科学计算、图像处理和数据分析等领域。在MATLAB中实现BP神经网络，可以...
浅谈sklearn中predict与predict_proba区别
09-16
在机器学习领域，特别是在使用Python的scikit-learn库时，`predict` 和 `predict_proba` 是两个非常重要的方法，它们用于从训练好的模型中获取预测结果。这两个方法在逻辑回归（LogisticRegression）以及其他分类...
Python sklearn中的.fit与.predict的用法说明
09-16
在Python的机器学习库scikit-learn（简称sklearn）中，`.fit`和`.predict`是两个核心方法，它们在构建和应用模型时起到关键作用。本文将深入讲解这两个方法以及相关的知识点。 首先，`.fit`方法是用于训练模型的。...
predict_image_预测_图像识别_predict_image_
10-02
"predict_image"通常指的是一个功能或脚本，用于执行这个任务。在这个场景中，我们主要讨论如何配置和实现图像的预测分类。 首先，我们需要理解图像识别的基本流程。图像识别包括图像预处理、特征提取、模型训练和...
chromedriver-mac-arm64_126.0.6474.0.zip
07-31
chromedriver-mac-arm64_126.0.6474.0.zip
chromedriver-mac-arm64_128.0.6548.0.zip
07-31
chromedriver-mac-arm64_128.0.6548.0.zip
MySQL查询加速器：利用Query Cache提升效率
07-31
MySQL是一个流行的开源关系型数据库管理系统（RDBMS），广泛用于Web应用程序的后端数据存储。它基于结构化查询语言（SQL）来管理数据，并且是LAMP（Linux, Apache, MySQL, PHP/Python/Perl）技术栈的一部分，这个技术栈常用于构建动态网站和Web应用程序。 MySQL的特点包括： - **开放源代码**：MySQL的源代码是公开的，任何人都可以自由使用和修改。 - **跨平台**：MySQL可以在多种操作系统上运行，包括Linux、Windows、macOS等。 - **高性能**：MySQL以其快速的查询处理和良好的性能而闻名。 - **可靠性**：MySQL提供了多种机制来确保数据的完整性和可靠性，包括事务支持、备份和恢复功能。 - **易于使用**：MySQL提供了简单直观的界面和丰富的文档，便于用户学习和使用。 - **可扩展性**：MySQL支持从小型应用到大型企业级应用的扩展。 - **社区支持**：由于其广泛的使用，MySQL拥有一个活跃的开发者社区，提供大量的资源和支持。 MySQL被广泛应用于各种场景，包括在线事务处理（OL
Objective-C语言的基础教程.md
最新发布
07-31
Objective-C是一种面向对象的编程语言，是C语言的扩展。它主要用于苹果的iOS和macOS应用程序开发。
Bike Sharing Demand数据集如何进行统计推断，给出详细代码
06-03
Bike Sharing Demand 数据集可以通过许多统计方法进行推断，例如线性回归、决策树、随机森林等。下面是一个使用线性回归模型进行推断的代码示例： ```python import pandas as pd import numpy as np from sklearn.linear_model import LinearRegression from sklearn.model_selection import train_test_split # 读取数据集 data = pd.read_csv('BikeSharingDemand.csv') # 数据预处理 data['datetime'] = pd.to_datetime(data['datetime']) data['year'] = data['datetime'].dt.year data['month'] = data['datetime'].dt.month data['day'] = data['datetime'].dt.day data['hour'] = data['datetime'].dt.hour # 特征选择 X = data[['temp', 'atemp', 'humidity', 'windspeed', 'year', 'month', 'day', 'hour']] y = data['count'] # 数据集划分 X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42) # 模型训练 model = LinearRegression() model.fit(X_train, y_train) # 模型评估 score = model.score(X_test, y_test) print('R2 score:', score) # 预测 y_pred = model.predict(X_test) ``` 这段代码的功能包括读取数据集、数据预处理、特征选择、数据集划分、模型训练、模型评估和预测。其中，特征选择包括选择了温度、湿度、风速等特征，数据集划分采用了 80/20 的比例，模型评估采用 R2 分数。你可以根据需要进行修改和优化。
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值