notebook.community

Edit and run



In [6]:

    
import pandas as pd
import numpy as np

raw_df = pd.DataFrame.from_csv('31461-0001-Data.tsv',sep='\t', index_col=None)



In [60]:









    Out[60]:





109



In [13]:

    
field = 'r111r'
field2 = 'r112r'
sa2 = raw_df[['newid','year','country','cow','r111r','r112r','r113r','r114r']]
sa2 = sa2[sa2.r111r != ' ']



In [21]:

    
trans = ["Algeria",
"Argentina",
"Belarus",
"Benin",
"Bolivia",
"Brazil",
"Burkina Faso",
"Cambodia",
"Cent. Af. Rep.",
"Chad",
"Congo",
"Cote d'Ivoire",
"El Salvador",
"Estonia",
"Ethiopia",
"Gabon",
"Georgia",
"Ghana",
"Guatemala",
"Guinea-Bissau",
"Honduras",
"Hungary",
"Indonesia",
"Madagascar",
"Mali",
"Moldova",
"Mongolia",
"Mozambique",
"Namibia",
"Nepal",
"Nicaragua",
"Niger",
"Nigeria",
"Pakistan",
"Paraguay",
"Peru",
"Philippines",
"Poland",
"Rwanda",
"Romania",
"Slovenia",
"Sudan",
"Suriname",
"Taiwan",
"Tajikistan",
"Turkey",
"Uruguay",
"Zambia",
"Zimbabwe"]



In [22]:

    
trans_df = sa2[sa2.country.isin(trans)]



In [23]:

    
trans_df









    Out[23]:






  
    
      
      newid
      year
      country
      cow
      r111r
      r112r
      r113r
      r114r
    
  
  
    
      17
      84
      1997
      Algeria
      615
      1
      1
      1
      1
    
    
      20
      130
      1983
      Argentina
      160
      0
      0
      1
      2
    
    
      60
      368
      1995
      Belarus
      370
      0
      0
      3
      0
    
    
      61
      368
      1995
      Belarus
      370
      0
      1
      3
      0
    
    
      62
      368
      1995
      Belarus
      370
      0
      0
      2
      0
    
    
      63
      369
      1995
      Belarus
      370
      0
      0
      2
      1
    
    
      64
      374
      2000
      Belarus
      370
      2
      2
      2
      1
    
    
      65
      374
      2000
      Belarus
      370
      3
      3
      3
      3
    
    
      66
      375
      2001
      Belarus
      370
      0
      0
      1
      3
    
    
      67
      375
      2001
      Belarus
      370
      3
      3
      3
      3
    
    
      68
      375
      2001
      Belarus
      370
      3
      1
      3
      3
    
    
      69
      375
      2001
      Belarus
      370
      1
      2
      2
      2
    
    
      70
      375
      2001
      Belarus
      370
      0
      0
      0
      0
    
    
      71
      378
      2004
      Belarus
      370
      0
      2
      1
      2
    
    
      72
      378
      2004
      Belarus
      370
      0
      0
      0
      0
    
    
      73
      455
      1996
      Benin
      434
      1
      0
      1
      1
    
    
      74
      516
      1997
      Bolivia
      145
      1
      0
      0
      0
    
    
      75
      521
      2002
      Bolivia
      145
      0
      0
      1
      0
    
    
      89
      586
      1994
      Brazil
      140
      0
      0
      1
      0
    
    
      103
      761
      1993
      Cambodia
      811
      2
      0
      0
      2
    
    
      104
      766
      1998
      Cambodia
      811
      1
      0
      2
      0
    
    
      105
      766
      1998
      Cambodia
      811
      0
      1
      2
      3
    
    
      106
      766
      1998
      Cambodia
      811
      0
      1
      2
      3
    
    
      107
      766
      1998
      Cambodia
      811
      0
      0
      1
      1
    
    
      108
      766
      1998
      Cambodia
      811
      0
      0
      2
      0
    
    
      109
      766
      1998
      Cambodia
      811
      0
      2
      3
      3
    
    
      110
      771
      2003
      Cambodia
      811
      1
      2
      3
      3
    
    
      111
      771
      2003
      Cambodia
      811
      3
      0
      2
      1
    
    
      112
      771
      2003
      Cambodia
      811
      3
      0
      2
      3
    
    
      123
      1022
      2002
      Congo
      484
      3
      0
      0
      0
    
    
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
    
    
      429
      3700
      1996
      Romania
      360
      0
      0
      1
      1
    
    
      430
      3700
      1996
      Romania
      360
      0
      0
      2
      0
    
    
      431
      3704
      2000
      Romania
      360
      0
      0
      1
      0
    
    
      432
      3708
      2004
      Romania
      360
      1
      0
      2
      1
    
    
      455
      3750
      2003
      Rwanda
      517
      3
      3
      1
      2
    
    
      456
      3750
      2003
      Rwanda
      517
      2
      3
      0
      3
    
    
      457
      3751
      2003
      Rwanda
      517
      1
      3
      0
      3
    
    
      498
      3945
      1992
      Slovenia
      349
      1
      0
      1
      0
    
    
      506
      4164
      1991
      Suriname
      115
      0
      1
      0
      1
    
    
      507
      4169
      1996
      Suriname
      115
      1
      0
      1
      1
    
    
      508
      4173
      2000
      Suriname
      115
      0
      0
      0
      0
    
    
      511
      4328
      2004
      Taiwan
      713
      2
      0
      1
      0
    
    
      512
      4338
      2000
      Tajikistan
      702
      0
      0
      2
      1
    
    
      529
      4555
      2002
      Turkey
      640
      0
      0
      0
      1
    
    
      530
      4555
      2002
      Turkey
      640
      0
      0
      0
      0
    
    
      565
      4719
      1984
      Uruguay
      165
      0
      1
      2
      1
    
    
      579
      4974
      1991
      Zambia
      551
      2
      1
      1
      1
    
    
      580
      4974
      1991
      Zambia
      551
      3
      1
      2
      0
    
    
      581
      4984
      2001
      Zambia
      551
      2
      0
      2
      1
    
    
      582
      4984
      2001
      Zambia
      551
      2
      0
      2
      0
    
    
      583
      4984
      2001
      Zambia
      551
      1
      1
      1
      0
    
    
      584
      4998
      1985
      Zimbabwe
      552
      2
      2
      2
      3
    
    
      585
      5013
      2000
      Zimbabwe
      552
      0
      0
      2
      3
    
    
      586
      5013
      2000
      Zimbabwe
      552
      2
      2
      2
      2
    
    
      587
      5013
      2000
      Zimbabwe
      552
      1
      3
      3
      3
    
    
      588
      5013
      2000
      Zimbabwe
      552
      2
      2
      1
      3
    
    
      589
      5013
      2000
      Zimbabwe
      552
      0
      0
      3
      3
    
    
      590
      5015
      2002
      Zimbabwe
      552
      0
      2
      3
      3
    
    
      591
      5015
      2002
      Zimbabwe
      552
      0
      3
      3
      3
    
    
      592
      5015
      2002
      Zimbabwe
      552
      1
      2
      2
      3
    
  

202 rows × 8 columns



In [57]:

    
for c in raw_df.country.unique():
    c_df = trans_df[trans_df.country == c]
    r111r = c_df.r111r.max()
    r112r = c_df.r112r.max()
    r113r = c_df.r113r.max()
    r114r = c_df.r114r.max()
    print "{0}, {1}, {2}, {3}, {4}".format(c,r111r, r112r, r113r, r114r)









    



Afghanistan, nan, nan, nan, nan
Albania, nan, nan, nan, nan
Algeria, 1, 1, 1, 1
Angola, nan, nan, nan, nan
Argentina, 0, 0, 1, 2
Armenia, nan, nan, nan, nan
Azerbaijan, nan, nan, nan, nan
Bangladesh, nan, nan, nan, nan
Belarus, 3, 3, 3, 3
Benin, 1, 0, 1, 1
Bolivia, 1, 0, 1, 0
Bosnia-Herz, nan, nan, nan, nan
Botswana, nan, nan, nan, nan
Brazil, 0, 0, 1, 0
Bulgaria, nan, nan, nan, nan
Burundi, nan, nan, nan, nan
Cambodia, 3, 2, 3, 3
Cameroon, nan, nan, nan, nan
Chile, nan, nan, nan, nan
Colombia, nan, nan, nan, nan
Congo, 3, 0, 0, 0
Costa Rica, nan, nan, nan, nan
Cote d'Ivoire, nan, nan, nan, nan
Croatia, nan, nan, nan, nan
Czech Rep., nan, nan, nan, nan
Dom. Rep., nan, nan, nan, nan
Ecuador, nan, nan, nan, nan
El Salvador, 1, 1, 2, 1
Eq. Guinea, nan, nan, nan, nan
Estonia, 0, 0, 1, 0
Fiji, nan, nan, nan, nan
France, nan, nan, nan, nan
GDR, nan, nan, nan, nan
Gambia, nan, nan, nan, nan
Georgia, 3, 2, 2, 2
Ghana, 2, 0, 1, 2
Grenada, nan, nan, nan, nan
Guatemala, 3, 2, 2, 3
Guinea, nan, nan, nan, nan
Guyana, nan, nan, nan, nan
Haiti, nan, nan, nan, nan
Honduras, 0, 0, 0, 1
Hungary, 1, 0, 1, 1
Indonesia, 3, 2, 1, 2
Jamaica, nan, nan, nan, nan
Kazakhstan, nan, nan, nan, nan
Kenya, nan, nan, nan, nan
Kyrgyzstan, nan, nan, nan, nan
Latvia, nan, nan, nan, nan
Lesotho, nan, nan, nan, nan
Liberia, nan, nan, nan, nan
Lithuania, nan, nan, nan, nan
Macedonia, nan, nan, nan, nan
Madagascar, 2, 0, 0, 2
Malawi, nan, nan, nan, nan
Malaysia, nan, nan, nan, nan
Mali, 0, 0, 0, 0
Mauritius, nan, nan, nan, nan
Mexico, nan, nan, nan, nan
Moldova, 1, 0, 1, 1
Mongolia, 3, 1, 1, 0
Morocco, nan, nan, nan, nan
Mozambique, 3, 1, 2, 2
Namibia, 2, 1, 2, 2
Nepal, 2, 0, 1, 1
Nicaragua, 2, 2, 2, 2
Niger, 3, 3, 3, 0
Nigeria, 2, 0, 1, 1
P. N. Guinea, nan, nan, nan, nan
Pakistan, 3, 3, 3, 3
Panama, nan, nan, nan, nan
Paraguay, 3, 0, 3, 1
Peru, 3, 1, 3, 3
Philippines, 2, 2, 2, 2
Poland, 0, 0, 0, 0
Romania, 3, 2, 3, 3
Russia, nan, nan, nan, nan
Rwanda, 3, 3, 1, 3
S. Africa, nan, nan, nan, nan
Senegal, nan, nan, nan, nan
Serbia, nan, nan, nan, nan
Montenegro, nan, nan, nan, nan
Sierra Leone, nan, nan, nan, nan
Slovakia, nan, nan, nan, nan
Slovenia, 1, 0, 1, 0
Solomon Is., nan, nan, nan, nan
Spain, nan, nan, nan, nan
Sri Lanka, nan, nan, nan, nan
Suriname, 1, 1, 1, 1
Swaziland, nan, nan, nan, nan
Taiwan, 2, 0, 1, 0
Tajikistan, 0, 0, 2, 1
Tanzania, nan, nan, nan, nan
Tanzania -Zanzibar, nan, nan, nan, nan
Thailand, nan, nan, nan, nan
Timor-Leste, nan, nan, nan, nan
Togo, nan, nan, nan, nan
Trinidad-Tobago, nan, nan, nan, nan
Tunisia, nan, nan, nan, nan
Turkey, 0, 0, 0, 1
USA, nan, nan, nan, nan
Uganda, nan, nan, nan, nan
Ukraine, nan, nan, nan, nan
Uruguay, 0, 1, 2, 1
Uzbekistan, nan, nan, nan, nan
Venezuela, nan, nan, nan, nan
Yemen, nan, nan, nan, nan
Zambia, 3, 1, 2, 1
Zimbabwe, 2, 3, 3, 3



In [42]:

    
trans_indicator_df = pd.DataFrame.from_csv('r111-r114_indicators_trans.csv')



In [55]:

    
print "r111r pct: {0}".format(len(trans_indicator_df[trans_indicator_df.r111r >= 2])/float(len(trans_indicator_df)))
print "r112r pct: {0}".format(len(trans_indicator_df[trans_indicator_df.r112r >= 2])/float(len(trans_indicator_df)))
print "r113r pct: {0}".format(len(trans_indicator_df[trans_indicator_df.r113r >= 2])/float(len(trans_indicator_df)))
print "r11rr pct: {0}".format(len(trans_indicator_df[trans_indicator_df.r114r >= 2])/float(len(trans_indicator_df)))









    



r111r pct: 0.585365853659
r112r pct: 0.292682926829
r113r pct: 0.439024390244
r11rr pct: 0.414634146341



In [63]:

    
all_indicator_df = pd.DataFrame.from_csv('r111-r114_indicators_all.csv')



In [64]:

    
print "r111r pct: {0}".format(len(all_indicator_df[all_indicator_df.r111r >= 2])/float(len(all_indicator_df)))
print "r112r pct: {0}".format(len(all_indicator_df[all_indicator_df.r112r >= 2])/float(len(all_indicator_df)))
print "r113r pct: {0}".format(len(all_indicator_df[all_indicator_df.r113r >= 2])/float(len(all_indicator_df)))
print "r11rr pct: {0}".format(len(all_indicator_df[all_indicator_df.r114r >= 2])/float(len(all_indicator_df)))









    



r111r pct: 0.220183486239
r112r pct: 0.110091743119
r113r pct: 0.165137614679
r11rr pct: 0.155963302752



In [ ]:

	newid	year	country	cow	r111r	r112r	r113r	r114r
17	84	1997	Algeria	615	1	1	1	1
20	130	1983	Argentina	160	0	0	1	2
60	368	1995	Belarus	370	0	0	3	0
61	368	1995	Belarus	370	0	1	3	0
62	368	1995	Belarus	370	0	0	2	0
63	369	1995	Belarus	370	0	0	2	1
64	374	2000	Belarus	370	2	2	2	1
65	374	2000	Belarus	370	3	3	3	3
66	375	2001	Belarus	370	0	0	1	3
67	375	2001	Belarus	370	3	3	3	3
68	375	2001	Belarus	370	3	1	3	3
69	375	2001	Belarus	370	1	2	2	2
70	375	2001	Belarus	370	0	0	0	0
71	378	2004	Belarus	370	0	2	1	2
72	378	2004	Belarus	370	0	0	0	0
73	455	1996	Benin	434	1	0	1	1
74	516	1997	Bolivia	145	1	0	0	0
75	521	2002	Bolivia	145	0	0	1	0
89	586	1994	Brazil	140	0	0	1	0
103	761	1993	Cambodia	811	2	0	0	2
104	766	1998	Cambodia	811	1	0	2	0
105	766	1998	Cambodia	811	0	1	2	3
106	766	1998	Cambodia	811	0	1	2	3
107	766	1998	Cambodia	811	0	0	1	1
108	766	1998	Cambodia	811	0	0	2	0
109	766	1998	Cambodia	811	0	2	3	3
110	771	2003	Cambodia	811	1	2	3	3
111	771	2003	Cambodia	811	3	0	2	1
112	771	2003	Cambodia	811	3	0	2	3
123	1022	2002	Congo	484	3	0	0	0
...	...	...	...	...	...	...	...	...
429	3700	1996	Romania	360	0	0	1	1
430	3700	1996	Romania	360	0	0	2	0
431	3704	2000	Romania	360	0	0	1	0
432	3708	2004	Romania	360	1	0	2	1
455	3750	2003	Rwanda	517	3	3	1	2
456	3750	2003	Rwanda	517	2	3	0	3
457	3751	2003	Rwanda	517	1	3	0	3
498	3945	1992	Slovenia	349	1	0	1	0
506	4164	1991	Suriname	115	0	1	0	1
507	4169	1996	Suriname	115	1	0	1	1
508	4173	2000	Suriname	115	0	0	0	0
511	4328	2004	Taiwan	713	2	0	1	0
512	4338	2000	Tajikistan	702	0	0	2	1
529	4555	2002	Turkey	640	0	0	0	1
530	4555	2002	Turkey	640	0	0	0	0
565	4719	1984	Uruguay	165	0	1	2	1
579	4974	1991	Zambia	551	2	1	1	1
580	4974	1991	Zambia	551	3	1	2	0
581	4984	2001	Zambia	551	2	0	2	1
582	4984	2001	Zambia	551	2	0	2	0
583	4984	2001	Zambia	551	1	1	1	0
584	4998	1985	Zimbabwe	552	2	2	2	3
585	5013	2000	Zimbabwe	552	0	0	2	3
586	5013	2000	Zimbabwe	552	2	2	2	2
587	5013	2000	Zimbabwe	552	1	3	3	3
588	5013	2000	Zimbabwe	552	2	2	1	3
589	5013	2000	Zimbabwe	552	0	0	3	3
590	5015	2002	Zimbabwe	552	0	2	3	3
591	5015	2002	Zimbabwe	552	0	3	3	3
592	5015	2002	Zimbabwe	552	1	2	2	3