【数据分析可视化】分组和透视功能实战-航班数据分析

数据来源
https://fivethirtyeight.com/politics/

import numpy as np
import pandas as pd
from pandas import Series, DataFrame

# 航班数据半个月20w条
link = '/Users/bennyrhys/Desktop/数据分析可视化-数据集/homework/usa_flights.csv'
df = pd.read_csv(link)
df.head()

	flight_date	unique_carrier	flight_num	origin	dest	arr_delay	cancelled	distance	carrier_delay	weather_delay	late_aircraft_delay	nas_delay	security_delay	actual_elapsed_time
0	02/01/2015 0:00	AA	1	JFK	LAX	-19.0	0	2475	NaN	NaN	NaN	NaN	NaN	381.0
1	03/01/2015 0:00	AA	1	JFK	LAX	-39.0	0	2475	NaN	NaN	NaN	NaN	NaN	358.0
2	04/01/2015 0:00	AA	1	JFK	LAX	-12.0	0	2475	NaN	NaN	NaN	NaN	NaN	385.0
3	05/01/2015 0:00	AA	1	JFK	LAX	-8.0	0	2475	NaN	NaN	NaN	NaN	NaN	389.0
4	06/01/2015 0:00	AA	1	JFK	LAX	25.0	0	2475	0.0	0.0	0.0	25.0	0.0	424.0

df.tail()

	flight_date	unique_carrier	flight_num	origin	dest	arr_delay	cancelled	distance	carrier_delay	weather_delay	late_aircraft_delay	nas_delay	security_delay	actual_elapsed_time
201659	10/01/2015 0:00	NK	188	OAK	LAS	-16.0	0	407	NaN	NaN	NaN	NaN	NaN	77.0
201660	11/01/2015 0:00	NK	188	OAK	LAS	-4.0	0	407	NaN	NaN	NaN	NaN	NaN	87.0
201661	12/01/2015 0:00	NK	188	OAK	LAS	-7.0	0	407	NaN	NaN	NaN	NaN	NaN	82.0
201662	13/01/2015 0:00	NK	188	OAK	LAS	23.0	0	407	3.0	0.0	0.0	20.0	0.0	103.0
201663	14/01/2015 0:00	NK	188	OAK	LAS	-7.0	0	407	NaN	NaN	NaN	NaN	NaN	82.0

# 行列信息
df.shape

(201664, 14)

判断延误arr_delay>0就是延误

排序航班到达时间，前十名递减

df.sort_values('arr_delay', ascending=False)[:10]

	flight_date	unique_carrier	flight_num	origin	dest	arr_delay	cancelled	distance	carrier_delay	weather_delay	late_aircraft_delay	nas_delay	security_delay	actual_elapsed_time
11073	11/01/2015 0:00	AA	1595	AUS	DFW	1444.0	0	190	1444.0	0.0	0.0	0.0	0.0	59.0
10214	13/01/2015 0:00	AA	1487	OMA	DFW	1392.0	0	583	1392.0	0.0	0.0	0.0	0.0	117.0
12430	03/01/2015 0:00	AA	1677	MEM	DFW	1384.0	0	432	1380.0	0.0	0.0	4.0	0.0	104.0
8443	04/01/2015 0:00	AA	1279	OMA	DFW	1237.0	0	583	1222.0	0.0	15.0	0.0	0.0	102.0
10328	05/01/2015 0:00	AA	1495	EGE	DFW	1187.0	0	721	1019.0	0.0	168.0	0.0	0.0	127.0
36570	04/01/2015 0:00	DL	1435	MIA	MSP	1174.0	0	1501	1174.0	0.0	0.0	0.0	0.0	231.0
36495	04/01/2015 0:00	DL	1367	ROC	ATL	1138.0	0	749	1112.0	0.0	0.0	26.0	0.0	171.0
59072	14/01/2015 0:00	DL	1687	SAN	MSP	1084.0	0	1532	1070.0	0.0	0.0	14.0	0.0	240.0
32173	05/01/2015 0:00	AA	970	LAS	LAX	1042.0	0	236	1033.0	0.0	9.0	0.0	0.0	66.0
56488	12/01/2015 0:00	DL	2117	ATL	COS	1016.0	0	1184	1016.0	0.0	0.0	0.0	0.0	193.0

计算延误和没有延误所占比例

df['cancelled'].value_counts()

0    196873
1      4791
Name: cancelled, dtype: int64

# 新增一列表示是否延误
df['delayed'] = df['arr_delay'].apply(lambda x: x > 0)
df.head()

	flight_date	unique_carrier	flight_num	origin	dest	arr_delay	cancelled	distance	carrier_delay	weather_delay	late_aircraft_delay	nas_delay	security_delay	actual_elapsed_time	delayed
0	02/01/2015 0:00	AA	1	JFK	LAX	-19.0	0	2475	NaN	NaN	NaN	NaN	NaN	381.0	False
1	03/01/2015 0:00	AA	1	JFK	LAX	-39.0	0	2475	NaN	NaN	NaN	NaN	NaN	358.0	False
2	04/01/2015 0:00	AA	1	JFK	LAX	-12.0	0	2475	NaN	NaN	NaN	NaN	NaN	385.0	False
3	05/01/2015 0:00	AA	1	JFK	LAX	-8.0	0	2475	NaN	NaN	NaN	NaN	NaN	389.0	False
4	06/01/2015 0:00	AA	1	JFK	LAX	25.0	0	2475	0.0	0.0	0.0	25.0	0.0	424.0	True

delay_data = df['delayed'].value_counts()
delay_data

False    103037
True      98627
Name: delayed, dtype: int64

type(delay_data)

pandas.core.series.Series

delay_data[1] / (delay_data[0] + delay_data[1])

0.4890659711202793

每一个航空公司延误的情况

delay_group = df.groupby(['unique_carrier', 'delayed'])
delay_group

<pandas.core.groupby.generic.DataFrameGroupBy object at 0x11ff50710>

# 这是多级Series类型
delay_group.size()

unique_carrier  delayed
AA              False       8912
                True        9841
AS              False       3527
                True        2104
B6              False       4832
                True        4401
DL              False      17719
                True        9803
EV              False      10596
                True       11371
F9              False       1103
                True        1848
HA              False       1351
                True        1354
MQ              False       4692
                True        8060
NK              False       1550
                True        2133
OO              False       9977
                True       10804
UA              False       7885
                True        8624
US              False       7850
                True        6353
VX              False       1254
                True         781
WN              False      21789
                True       21150
dtype: int64

# 将多级Series转换为DataFrame
df_delay = delay_group.size().unstack()
df_delay

delayed	False	True
unique_carrier
AA	8912	9841
AS	3527	2104
B6	4832	4401
DL	17719	9803
EV	10596	11371
F9	1103	1848
HA	1351	1354
MQ	4692	8060
NK	1550	2133
OO	9977	10804
UA	7885	8624
US	7850	6353
VX	1254	781
WN	21789	21150

# 画图
import matplotlib.pyplot as plt

df_delay.plot()

<matplotlib.axes._subplots.AxesSubplot at 0x1210efb50>

plt.show()

# 蓝色没有延误
df_delay.plot(kind='barh', stacked=True, figsize=[16,6], colormap='winter')

<matplotlib.axes._subplots.AxesSubplot at 0x11c9e2290>

透视表功能

flights_by_carrier = df.pivot_table(index='flight_date', columns='unique_carrier')
flights_by_carrier.head()

	actual_elapsed_time										...	weather_delay
unique_carrier	AA	AS	B6	DL	EV	F9	HA	MQ	NK	OO	...	EV	F9	HA	MQ	NK	OO	UA	US	VX	WN
flight_date
02/01/2015 0:00	176.852122	182.872117	174.580475	155.151703	103.325014	148.674603	108.654709	99.608205	156.775439	101.953434	...	0.214118	0.000000	7.621849	4.956916	0.0	1.154150	0.816867	0.131429	4.413793	0.674080
03/01/2015 0:00	177.679298	189.126126	178.595474	161.668481	107.364508	158.116667	108.158416	102.381295	162.085106	107.836902	...	1.605061	0.454545	0.000000	4.263838	0.0	1.369444	1.317901	2.938053	5.350000	1.442974
04/01/2015 0:00	178.200938	184.766376	179.517287	156.963620	104.893505	149.746888	104.878641	109.936095	162.289753	106.465820	...	1.160754	0.374269	0.000000	9.286834	0.0	1.006859	2.996965	1.350000	6.414634	1.116999
05/01/2015 0:00	176.660858	178.226328	174.088000	142.719584	103.107938	152.104839	103.091787	108.128505	164.081560	103.141727	...	4.005384	0.586957	0.000000	12.048822	0.0	2.732057	8.422122	0.116838	3.312500	1.370968
06/01/2015 0:00	171.155474	173.200483	175.029326	144.458049	100.694926	146.739837	100.168317	100.343423	165.410909	99.287270	...	2.685092	4.866667	2.666667	7.971370	0.0	2.753521	6.551102	6.155660	13.615385	2.797213

5 rows × 154 columns