Group Partners: Filis, Manisha, Pablo
Ran the following commands in terminal to install Profiling (visualizing data), and Altair (map):
conda install -c conda-forge altair vega_datasets notebook vega
conda install -c anaconda pandas-profiling
#Data Manipulation
import pandas as pd
import numpy as np
# #Making Map Visualizations
# import altair as alt
# alt.renderers.enable('notebook')
# from vega_datasets import data
# #Making Line Plot Visualizations
# import plotly.plotly as py
# import cufflinks as cf
# from plotly.offline import download_plotlyjs, init_notebook_mode, plot, iplot
# init_notebook_mode(connected=True)
# cf.go_offline()
#Displaying EDA Profile
import pandas_profiling
#Disabling warnings
import warnings
warnings.filterwarnings('ignore')
# Classifiers
from sklearn.ensemble import BaggingClassifier, RandomForestClassifier
from random import randint
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn import tree
from sklearn import svm
# Measuring models and feature importance
from sklearn import metrics
from sklearn.metrics import accuracy_score, confusion_matrix, classification_report, roc_curve, auc
import matplotlib.pyplot as plt
%matplotlib inline
# sklearn processing stuff
from sklearn.model_selection import GridSearchCV
from sklearn.pipeline import make_pipeline
from sklearn.model_selection import cross_validate, cross_val_score, cross_val_predict
from sklearn import preprocessingdf = pd.read_csv('Life Expectancy Data.csv')
df.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Country | Year | Status | Life expectancy | Adult Mortality | infant deaths | Alcohol | percentage expenditure | Hepatitis B | Measles | ... | Polio | Total expenditure | Diphtheria | HIV/AIDS | GDP | Population | thinness 1-19 years | thinness 5-9 years | Income composition of resources | Schooling | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Afghanistan | 2015 | Developing | 65.0 | 263.0 | 62 | 0.01 | 71.279624 | 65.0 | 1154 | ... | 6.0 | 8.16 | 65.0 | 0.1 | 584.259210 | 33736494.0 | 17.2 | 17.3 | 0.479 | 10.1 |
| 1 | Afghanistan | 2014 | Developing | 59.9 | 271.0 | 64 | 0.01 | 73.523582 | 62.0 | 492 | ... | 58.0 | 8.18 | 62.0 | 0.1 | 612.696514 | 327582.0 | 17.5 | 17.5 | 0.476 | 10.0 |
| 2 | Afghanistan | 2013 | Developing | 59.9 | 268.0 | 66 | 0.01 | 73.219243 | 64.0 | 430 | ... | 62.0 | 8.13 | 64.0 | 0.1 | 631.744976 | 31731688.0 | 17.7 | 17.7 | 0.470 | 9.9 |
| 3 | Afghanistan | 2012 | Developing | 59.5 | 272.0 | 69 | 0.01 | 78.184215 | 67.0 | 2787 | ... | 67.0 | 8.52 | 67.0 | 0.1 | 669.959000 | 3696958.0 | 17.9 | 18.0 | 0.463 | 9.8 |
| 4 | Afghanistan | 2011 | Developing | 59.2 | 275.0 | 71 | 0.01 | 7.097109 | 68.0 | 3013 | ... | 68.0 | 7.87 | 68.0 | 0.1 | 63.537231 | 2978599.0 | 18.2 | 18.2 | 0.454 | 9.5 |
5 rows × 22 columns
df.shape(2938, 22)
df.isna().sum().sort_values(ascending=False)Population 652
Hepatitis B 553
GDP 448
Total expenditure 226
Alcohol 194
Income composition of resources 167
Schooling 163
BMI 34
thinness 1-19 years 34
thinness 5-9 years 34
Diphtheria 19
Polio 19
Adult Mortality 10
Life expectancy 10
under-five deaths 0
HIV/AIDS 0
Measles 0
percentage expenditure 0
infant deaths 0
Status 0
Year 0
Country 0
dtype: int64
df['Life expectancy '].hist()<matplotlib.axes._subplots.AxesSubplot at 0x1a1efad7b8>
# #This plot of life expectancies is of one country with respect to years.
# df_group.loc['Albania'].iplot(y='Life expectancy ')
# print('Albania')df_group = df.set_index(['Country', 'Year'])
plt.rc('xtick',labelsize=20)
plt.rc('ytick',labelsize=20)
countries = df['Country'].unique()
fig = plt.figure(figsize=(14,12))
ax = fig.add_subplot(111)
plt.ylabel('Life Expectancy', fontsize=20)
plt.xlabel('Years', fontsize=20)
plt.title('Life Expectancies per Year', fontsize=20)
for country in countries[:20]:
df_group.loc[country].plot(y='Life expectancy ', ax=ax)
L = plt.legend(countries[:20])
plt.savefig('Life Expectacies per Year.png')
Finding: Most countries appear relatively flat, indicating a small correlation with time dependence. We investigate this more below.
# profile = pandas_profiling.ProfileReport(df)# from IPython.core.display import display, HTML
# display(HTML(profile.html))df.columnsIndex(['Country', 'Year', 'Status', 'Life expectancy ', 'Adult Mortality',
'infant deaths', 'Alcohol', 'percentage expenditure', 'Hepatitis B',
'Measles ', ' BMI ', 'under-five deaths ', 'Polio', 'Total expenditure',
'Diphtheria ', ' HIV/AIDS', 'GDP', 'Population',
' thinness 1-19 years', ' thinness 5-9 years',
'Income composition of resources', 'Schooling'],
dtype='object')
Status
Developed or Developing status
Life expectancy
Life Expectancy in age
Adult Mortality
Adult Mortality Rates of both sexes (probability of dying between 15 and 60 years per 1000 population)
infant deaths
Number of Infant Deaths per 1000 population
Alcohol
Alcohol, recorded per capita (15+) consumption (in litres of pure alcohol)
percentage expenditure
Expenditure on health as a percentage of Gross Domestic Product per capita(%)
Hepatitis B
Hepatitis B (HepB) immunization coverage among 1-year-olds (%)
Measles
Measles - number of reported cases per 1000 population
BMI
Average Body Mass Index of entire population
under-five deaths
Number of under-five deaths per 1000 population
Polio
Polio (Pol3) immunization coverage among 1-year-olds (%)
Total expenditure
General government expenditure on health as a percentage of total government expenditure (%)
Diphtheria
Diphtheria tetanus toxoid and pertussis (DTP3) immunization coverage among 1-year-olds (%)
HIV/AIDS
Deaths per 1 000 live births HIV/AIDS (0-4 years)
GDP
Gross Domestic Product per capita (in USD)
Population
Population of the country
thinness 1-19 years
Prevalence of thinness among children and adolescents for Age 10 to 19 (% )
thinness 5-9 years
Prevalence of thinness among children for Age 5 to 9(%)
Income composition of resources
Human Development Index in terms of income composition of resources (index ranging from 0 to 1) url: http://hdr.undp.org/en/content/human-development-index-hdi
Schooling
Number of years of Schooling(years)
df.corr()[1:2]
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Year | Life expectancy | Adult Mortality | infant deaths | Alcohol | percentage expenditure | Hepatitis B | Measles | BMI | under-five deaths | Polio | Total expenditure | Diphtheria | HIV/AIDS | GDP | Population | thinness 1-19 years | thinness 5-9 years | Income composition of resources | Schooling | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Life expectancy | 0.170033 | 1.0 | -0.696359 | -0.196557 | 0.404877 | 0.381864 | 0.256762 | -0.157586 | 0.567694 | -0.222529 | 0.465556 | 0.218086 | 0.479495 | -0.556556 | 0.461455 | -0.021538 | -0.477183 | -0.471584 | 0.724776 | 0.751975 |
print('Maximum year is', df['Year'].max())
print('Minimum year is', df['Year'].min())Maximum year is 2015
Minimum year is 2000
Let's filter by 2015 to make a map of life expectency per country.
df_2015 = df[df['Year']==2015]Reading in the country codes tsv to make the map (file generated from UN stats webpage: https://unstats.un.org/unsd/methodology/m49/
country_codes = pd.read_csv('country_code_2.tsv',sep='\t')
country_codes.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Country or Area | M49 code | ISO-alpha3 code | |
|---|---|---|---|
| 0 | Afghanistan | 4 | AFG |
| 1 | Aland Islands | 248 | ALA |
| 2 | Albania | 8 | ALB |
| 3 | Algeria | 12 | DZA |
| 4 | American Samoa | 16 | ASM |
Matching countries to country names in country codes tsv data
df_2015.replace("Côte d'Ivoire", "Cote d'Ivoire", inplace=True)
df_2015.replace('Swaziland', 'Eswatini', inplace=True)
df_2015.replace('The former Yugoslav republic of Macedonia', 'North Macedonia', inplace=True)
df_2015_map = df_2015.merge(country_codes, how="left", left_on="Country",right_on="Country or Area")
df_2015_map.rename(columns={'M49 code': 'id'}, inplace=True)
df_2015_map.drop(columns=['ISO-alpha3 code', 'Country or Area'], inplace=True)
df_2015_map['tooltip']=df_2015_map['Country']+': '+df_2015_map['Life expectancy '].astype(str) + ' yr'# source = alt.topo_feature(data.world_110m.url,'countries')
# map_plot = alt.Chart(source).mark_geoshape().encode(
# color=alt.Color('Life expectancy :Q', legend=alt.Legend(title='Years')),
# tooltip='tooltip:N'
# ).transform_lookup(
# lookup='id',
# from_=alt.LookupData(df_2015_map, 'id', ['Life expectancy ', 'tooltip'])
# ).project(
# type='equirectangular'
# ).properties(
# width=900,
# height=540,
# title=('Life Expectency in Years')
# )# map_plotFinding: The country with the highest life-expectancy is Slovenia (Europe) at 88 years, and the lowest is Sierra Leonne (Africa) at 52 years.
df_group = df.set_index(['Country','Year'])
df_group.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Status | Life expectancy | Adult Mortality | infant deaths | Alcohol | percentage expenditure | Hepatitis B | Measles | BMI | under-five deaths | Polio | Total expenditure | Diphtheria | HIV/AIDS | GDP | Population | thinness 1-19 years | thinness 5-9 years | Income composition of resources | Schooling | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Country | Year | ||||||||||||||||||||
| Afghanistan | 2015 | Developing | 65.0 | 263.0 | 62 | 0.01 | 71.279624 | 65.0 | 1154 | 19.1 | 83 | 6.0 | 8.16 | 65.0 | 0.1 | 584.259210 | 33736494.0 | 17.2 | 17.3 | 0.479 | 10.1 |
| 2014 | Developing | 59.9 | 271.0 | 64 | 0.01 | 73.523582 | 62.0 | 492 | 18.6 | 86 | 58.0 | 8.18 | 62.0 | 0.1 | 612.696514 | 327582.0 | 17.5 | 17.5 | 0.476 | 10.0 | |
| 2013 | Developing | 59.9 | 268.0 | 66 | 0.01 | 73.219243 | 64.0 | 430 | 18.1 | 89 | 62.0 | 8.13 | 64.0 | 0.1 | 631.744976 | 31731688.0 | 17.7 | 17.7 | 0.470 | 9.9 | |
| 2012 | Developing | 59.5 | 272.0 | 69 | 0.01 | 78.184215 | 67.0 | 2787 | 17.6 | 93 | 67.0 | 8.52 | 67.0 | 0.1 | 669.959000 | 3696958.0 | 17.9 | 18.0 | 0.463 | 9.8 | |
| 2011 | Developing | 59.2 | 275.0 | 71 | 0.01 | 7.097109 | 68.0 | 3013 | 17.2 | 97 | 68.0 | 7.87 | 68.0 | 0.1 | 63.537231 | 2978599.0 | 18.2 | 18.2 | 0.454 | 9.5 |
df_nona = df.dropna()df_nona.shape(1649, 22)
df_nona.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Country | Year | Status | Life expectancy | Adult Mortality | infant deaths | Alcohol | percentage expenditure | Hepatitis B | Measles | ... | Polio | Total expenditure | Diphtheria | HIV/AIDS | GDP | Population | thinness 1-19 years | thinness 5-9 years | Income composition of resources | Schooling | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Afghanistan | 2015 | Developing | 65.0 | 263.0 | 62 | 0.01 | 71.279624 | 65.0 | 1154 | ... | 6.0 | 8.16 | 65.0 | 0.1 | 584.259210 | 33736494.0 | 17.2 | 17.3 | 0.479 | 10.1 |
| 1 | Afghanistan | 2014 | Developing | 59.9 | 271.0 | 64 | 0.01 | 73.523582 | 62.0 | 492 | ... | 58.0 | 8.18 | 62.0 | 0.1 | 612.696514 | 327582.0 | 17.5 | 17.5 | 0.476 | 10.0 |
| 2 | Afghanistan | 2013 | Developing | 59.9 | 268.0 | 66 | 0.01 | 73.219243 | 64.0 | 430 | ... | 62.0 | 8.13 | 64.0 | 0.1 | 631.744976 | 31731688.0 | 17.7 | 17.7 | 0.470 | 9.9 |
| 3 | Afghanistan | 2012 | Developing | 59.5 | 272.0 | 69 | 0.01 | 78.184215 | 67.0 | 2787 | ... | 67.0 | 8.52 | 67.0 | 0.1 | 669.959000 | 3696958.0 | 17.9 | 18.0 | 0.463 | 9.8 |
| 4 | Afghanistan | 2011 | Developing | 59.2 | 275.0 | 71 | 0.01 | 7.097109 | 68.0 | 3013 | ... | 68.0 | 7.87 | 68.0 | 0.1 | 63.537231 | 2978599.0 | 18.2 | 18.2 | 0.454 | 9.5 |
5 rows × 22 columns
df_nona.rename(columns={'Life expectancy ':'target'}, inplace=True)df_nona.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Country | Year | Status | target | Adult Mortality | infant deaths | Alcohol | percentage expenditure | Hepatitis B | Measles | ... | Polio | Total expenditure | Diphtheria | HIV/AIDS | GDP | Population | thinness 1-19 years | thinness 5-9 years | Income composition of resources | Schooling | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Afghanistan | 2015 | Developing | 65.0 | 263.0 | 62 | 0.01 | 71.279624 | 65.0 | 1154 | ... | 6.0 | 8.16 | 65.0 | 0.1 | 584.259210 | 33736494.0 | 17.2 | 17.3 | 0.479 | 10.1 |
| 1 | Afghanistan | 2014 | Developing | 59.9 | 271.0 | 64 | 0.01 | 73.523582 | 62.0 | 492 | ... | 58.0 | 8.18 | 62.0 | 0.1 | 612.696514 | 327582.0 | 17.5 | 17.5 | 0.476 | 10.0 |
| 2 | Afghanistan | 2013 | Developing | 59.9 | 268.0 | 66 | 0.01 | 73.219243 | 64.0 | 430 | ... | 62.0 | 8.13 | 64.0 | 0.1 | 631.744976 | 31731688.0 | 17.7 | 17.7 | 0.470 | 9.9 |
| 3 | Afghanistan | 2012 | Developing | 59.5 | 272.0 | 69 | 0.01 | 78.184215 | 67.0 | 2787 | ... | 67.0 | 8.52 | 67.0 | 0.1 | 669.959000 | 3696958.0 | 17.9 | 18.0 | 0.463 | 9.8 |
| 4 | Afghanistan | 2011 | Developing | 59.2 | 275.0 | 71 | 0.01 | 7.097109 | 68.0 | 3013 | ... | 68.0 | 7.87 | 68.0 | 0.1 | 63.537231 | 2978599.0 | 18.2 | 18.2 | 0.454 | 9.5 |
5 rows × 22 columns
Spliting the data at the median value, and anything a above that age will be 1 for long life expectancy and 0 for low life expectancy.
df_nona['target'].describe()count 1649.000000
mean 69.302304
std 8.796834
min 44.000000
25% 64.400000
50% 71.700000
75% 75.000000
max 89.000000
Name: target, dtype: float64
median_age = np.median(df_nona['target'])
df_nona['target'] = df_nona.target.apply(lambda x: 1 if x >= median_age else 0)df_nona.target.value_counts()1 828
0 821
Name: target, dtype: int64
df_nona.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Country | Year | Status | target | Adult Mortality | infant deaths | Alcohol | percentage expenditure | Hepatitis B | Measles | ... | Polio | Total expenditure | Diphtheria | HIV/AIDS | GDP | Population | thinness 1-19 years | thinness 5-9 years | Income composition of resources | Schooling | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Afghanistan | 2015 | Developing | 0 | 263.0 | 62 | 0.01 | 71.279624 | 65.0 | 1154 | ... | 6.0 | 8.16 | 65.0 | 0.1 | 584.259210 | 33736494.0 | 17.2 | 17.3 | 0.479 | 10.1 |
| 1 | Afghanistan | 2014 | Developing | 0 | 271.0 | 64 | 0.01 | 73.523582 | 62.0 | 492 | ... | 58.0 | 8.18 | 62.0 | 0.1 | 612.696514 | 327582.0 | 17.5 | 17.5 | 0.476 | 10.0 |
| 2 | Afghanistan | 2013 | Developing | 0 | 268.0 | 66 | 0.01 | 73.219243 | 64.0 | 430 | ... | 62.0 | 8.13 | 64.0 | 0.1 | 631.744976 | 31731688.0 | 17.7 | 17.7 | 0.470 | 9.9 |
| 3 | Afghanistan | 2012 | Developing | 0 | 272.0 | 69 | 0.01 | 78.184215 | 67.0 | 2787 | ... | 67.0 | 8.52 | 67.0 | 0.1 | 669.959000 | 3696958.0 | 17.9 | 18.0 | 0.463 | 9.8 |
| 4 | Afghanistan | 2011 | Developing | 0 | 275.0 | 71 | 0.01 | 7.097109 | 68.0 | 3013 | ... | 68.0 | 7.87 | 68.0 | 0.1 | 63.537231 | 2978599.0 | 18.2 | 18.2 | 0.454 | 9.5 |
5 rows × 22 columns
# Count records the number of countries which are stationary
# The countries printed below (20% of all of our countries in total) may not be stationary -- the life expectancy changed
count = 0
for country in df_nona['Country'].unique() :
check = df_nona.loc[df_nona['Country']== country, 'target'].nunique()
if (check == 1):
count +=1
else:
print(country)Azerbaijan
Bangladesh
Belarus
Brazil
Bulgaria
Cabo Verde
Colombia
Dominican Republic
El Salvador
Guatemala
Honduras
Iraq
Latvia
Lithuania
Maldives
Mauritius
Morocco
Nicaragua
Romania
Russian Federation
Samoa
Sri Lanka
Suriname
Thailand
Tonga
Trinidad and Tobago
Turkey
Ukraine
Vanuatu
count / df_nona['Country'].nunique()0.7819548872180451
For 80% of the countries, life-expectency remained stationary. 20% may be non-stationary but overall have a small effect. Thus the life-expectancy does not depend on year.
#developed = 0, developing =1
#look into dropping one
df_nona = pd.concat([df_nona, pd.get_dummies(df_nona['Status'])], axis=1)
df_nona.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Country | Year | Status | target | Adult Mortality | infant deaths | Alcohol | percentage expenditure | Hepatitis B | Measles | ... | Diphtheria | HIV/AIDS | GDP | Population | thinness 1-19 years | thinness 5-9 years | Income composition of resources | Schooling | Developed | Developing | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Afghanistan | 2015 | Developing | 0 | 263.0 | 62 | 0.01 | 71.279624 | 65.0 | 1154 | ... | 65.0 | 0.1 | 584.259210 | 33736494.0 | 17.2 | 17.3 | 0.479 | 10.1 | 0 | 1 |
| 1 | Afghanistan | 2014 | Developing | 0 | 271.0 | 64 | 0.01 | 73.523582 | 62.0 | 492 | ... | 62.0 | 0.1 | 612.696514 | 327582.0 | 17.5 | 17.5 | 0.476 | 10.0 | 0 | 1 |
| 2 | Afghanistan | 2013 | Developing | 0 | 268.0 | 66 | 0.01 | 73.219243 | 64.0 | 430 | ... | 64.0 | 0.1 | 631.744976 | 31731688.0 | 17.7 | 17.7 | 0.470 | 9.9 | 0 | 1 |
| 3 | Afghanistan | 2012 | Developing | 0 | 272.0 | 69 | 0.01 | 78.184215 | 67.0 | 2787 | ... | 67.0 | 0.1 | 669.959000 | 3696958.0 | 17.9 | 18.0 | 0.463 | 9.8 | 0 | 1 |
| 4 | Afghanistan | 2011 | Developing | 0 | 275.0 | 71 | 0.01 | 7.097109 | 68.0 | 3013 | ... | 68.0 | 0.1 | 63.537231 | 2978599.0 | 18.2 | 18.2 | 0.454 | 9.5 | 0 | 1 |
5 rows × 24 columns
df_nona.drop(['Status'], axis=1, inplace=True)
df_nona.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Country | Year | target | Adult Mortality | infant deaths | Alcohol | percentage expenditure | Hepatitis B | Measles | BMI | ... | Diphtheria | HIV/AIDS | GDP | Population | thinness 1-19 years | thinness 5-9 years | Income composition of resources | Schooling | Developed | Developing | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Afghanistan | 2015 | 0 | 263.0 | 62 | 0.01 | 71.279624 | 65.0 | 1154 | 19.1 | ... | 65.0 | 0.1 | 584.259210 | 33736494.0 | 17.2 | 17.3 | 0.479 | 10.1 | 0 | 1 |
| 1 | Afghanistan | 2014 | 0 | 271.0 | 64 | 0.01 | 73.523582 | 62.0 | 492 | 18.6 | ... | 62.0 | 0.1 | 612.696514 | 327582.0 | 17.5 | 17.5 | 0.476 | 10.0 | 0 | 1 |
| 2 | Afghanistan | 2013 | 0 | 268.0 | 66 | 0.01 | 73.219243 | 64.0 | 430 | 18.1 | ... | 64.0 | 0.1 | 631.744976 | 31731688.0 | 17.7 | 17.7 | 0.470 | 9.9 | 0 | 1 |
| 3 | Afghanistan | 2012 | 0 | 272.0 | 69 | 0.01 | 78.184215 | 67.0 | 2787 | 17.6 | ... | 67.0 | 0.1 | 669.959000 | 3696958.0 | 17.9 | 18.0 | 0.463 | 9.8 | 0 | 1 |
| 4 | Afghanistan | 2011 | 0 | 275.0 | 71 | 0.01 | 7.097109 | 68.0 | 3013 | 17.2 | ... | 68.0 | 0.1 | 63.537231 | 2978599.0 | 18.2 | 18.2 | 0.454 | 9.5 | 0 | 1 |
5 rows × 23 columns
all_countries = df.Country.unique()
# Choose random countries
num_countries_to_withhold = 100
withhold_countries = set([])
while len(withhold_countries) < num_countries_to_withhold:
# Get a random integer so we can choose a random country
withhold_countries.add(all_countries[randint(0, len(all_countries) - 1)])
withhold_countries = list(withhold_countries)
training_countries = [country for country in all_countries if (country not in withhold_countries)]
df_country_split_a = df_nona[df_nona['Country'].isin(training_countries)]
df_country_split_b = df_nona[df_nona['Country'].isin(withhold_countries)]
num_countries_to_withhold/len(all_countries)0.5181347150259067
num_countries_to_withhold/len(all_countries)0.5181347150259067
We only train the model on 50% of our countries to check whether the country has an effect on life-expectency.
We train on group a and predict on group b
y = df_country_split_a['target']
X = df_country_split_a.drop(['target'], axis=1)X.drop(['Country'], axis=1, inplace = True)X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=1)classifier = DecisionTreeClassifier(random_state=10)
classifier.fit(X_train, y_train) DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=None,
max_features=None, 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, presort=False,
random_state=10, splitter='best')
y_pred = classifier.predict(X_test) acc = accuracy_score(y_test,y_pred) * 100
print("Accuracy is :{0}".format(acc))
# Check the AUC for predictions
false_positive_rate, true_positive_rate, thresholds = roc_curve(y_test, y_pred)
roc_auc = auc(false_positive_rate, true_positive_rate)
print("\nAUC is :{0}".format(round(roc_auc,2)))
# Create and print a confusion matrix
print('\nConfusion Matrix')
print('----------------')
pd.crosstab(y_test, y_pred, rownames=['True'], colnames=['Predicted'], margins=True)Accuracy is :91.53439153439153
AUC is :0.92
Confusion Matrix
----------------
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Predicted | 0 | 1 | All |
|---|---|---|---|
| True | |||
| 0 | 91 | 10 | 101 |
| 1 | 6 | 82 | 88 |
| All | 97 | 92 | 189 |
# The model misclassified 33 rowsclassifier.feature_importances_array([0.00409801, 0.18794948, 0. , 0.02129373, 0. ,
0. , 0.02089555, 0.05673548, 0. , 0. ,
0. , 0.00901852, 0.00666215, 0.01774478, 0.00532343,
0. , 0.00608392, 0.63835157, 0.02584337, 0. ,
0. ])
classifier.feature_importances_
def plot_feature_importances(model):
n_features = X_train.shape[1]
plt.figure(figsize=(8,8))
plt.barh(range(n_features), model.feature_importances_, align='center')
plt.yticks(np.arange(n_features), X_train.columns.values)
plt.xlabel("Feature importance")
plt.ylabel("Feature")
plot_feature_importances(classifier)
The Income composition of resources which is the Human Development Index combines education, life expectency, and GNI index (economic factor).
without Income composition of resources and Adult Mortality
X.drop(['Adult Mortality','Income composition of resources','Year'], axis=1, inplace=True)---------------------------------------------------------------------------
KeyError Traceback (most recent call last)
<ipython-input-110-d09afbee304e> in <module>
----> 1 X.drop(['Adult Mortality','Income composition of resources','Year','Developing'], axis=1, inplace=True)
~/anaconda3/lib/python3.7/site-packages/pandas/core/frame.py in drop(self, labels, axis, index, columns, level, inplace, errors)
3938 index=index, columns=columns,
3939 level=level, inplace=inplace,
-> 3940 errors=errors)
3941
3942 @rewrite_axis_style_signature('mapper', [('copy', True),
~/anaconda3/lib/python3.7/site-packages/pandas/core/generic.py in drop(self, labels, axis, index, columns, level, inplace, errors)
3778 for axis, labels in axes.items():
3779 if labels is not None:
-> 3780 obj = obj._drop_axis(labels, axis, level=level, errors=errors)
3781
3782 if inplace:
~/anaconda3/lib/python3.7/site-packages/pandas/core/generic.py in _drop_axis(self, labels, axis, level, errors)
3810 new_axis = axis.drop(labels, level=level, errors=errors)
3811 else:
-> 3812 new_axis = axis.drop(labels, errors=errors)
3813 result = self.reindex(**{axis_name: new_axis})
3814
~/anaconda3/lib/python3.7/site-packages/pandas/core/indexes/base.py in drop(self, labels, errors)
4963 if errors != 'ignore':
4964 raise KeyError(
-> 4965 '{} not found in axis'.format(labels[mask]))
4966 indexer = indexer[~mask]
4967 return self.delete(indexer)
KeyError: "['Adult Mortality' 'Income composition of resources' 'Year'] not found in axis"
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=1)forest=RandomForestClassifier(random_state=20)
param_grid = {
'n_estimators': [20,40,60,80,100],
'min_samples_split': [2, 4, 6],
'min_samples_leaf': [2, 4, 6]
}
gs = GridSearchCV(forest, param_grid, cv=5, return_train_score=True)gs.fit(X_train, y_train)GridSearchCV(cv=5, error_score='raise-deprecating',
estimator=RandomForestClassifier(bootstrap=True, class_weight=None,
criterion='gini', 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='warn', n_jobs=None,
oob_score=False, random_state=20,
verbose=0, warm_start=False),
iid='warn', n_jobs=None,
param_grid={'min_samples_leaf': [2, 4, 6],
'min_samples_split': [2, 4, 6],
'n_estimators': [20, 40, 60, 80, 100]},
pre_dispatch='2*n_jobs', refit=True, return_train_score=True,
scoring=None, verbose=0)
gs.best_estimator_RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',
max_depth=None, max_features='auto', max_leaf_nodes=None,
min_impurity_decrease=0.0, min_impurity_split=None,
min_samples_leaf=2, min_samples_split=2,
min_weight_fraction_leaf=0.0, n_estimators=100,
n_jobs=None, oob_score=False, random_state=20, verbose=0,
warm_start=False)
gs.best_params_{'min_samples_leaf': 2, 'min_samples_split': 2, 'n_estimators': 100}
#best mean test score across different combinations of hyperparameters
gs.best_score_0.9485815602836879
gs_cv_df=pd.DataFrame(gs.cv_results_)
gs_cv_df=gs_cv_df[gs_cv_df['params']=={'min_samples_leaf': 2, 'min_samples_split': 2, 'n_estimators': 100}]
gs_cv_df.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| mean_fit_time | std_fit_time | mean_score_time | std_score_time | param_min_samples_leaf | param_min_samples_split | param_n_estimators | params | split0_test_score | split1_test_score | ... | mean_test_score | std_test_score | rank_test_score | split0_train_score | split1_train_score | split2_train_score | split3_train_score | split4_train_score | mean_train_score | std_train_score | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 4 | 0.14169 | 0.02434 | 0.012077 | 0.00238 | 2 | 2 | 100 | {'min_samples_leaf': 2, 'min_samples_split': 2... | 0.938053 | 0.946903 | ... | 0.948582 | 0.020478 | 1 | 0.993348 | 0.997783 | 0.995565 | 0.993348 | 0.995575 | 0.995124 | 0.00166 |
1 rows × 23 columns
gs_cv_df=pd.DataFrame(gs.cv_results_)
gs_cv_df=gs_cv_df[gs_cv_df['rank_test_score']==1]
gs_cv_df.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| mean_fit_time | std_fit_time | mean_score_time | std_score_time | param_min_samples_leaf | param_min_samples_split | param_n_estimators | params | split0_test_score | split1_test_score | ... | mean_test_score | std_test_score | rank_test_score | split0_train_score | split1_train_score | split2_train_score | split3_train_score | split4_train_score | mean_train_score | std_train_score | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 4 | 0.141690 | 0.024340 | 0.012077 | 0.002380 | 2 | 2 | 100 | {'min_samples_leaf': 2, 'min_samples_split': 2... | 0.938053 | 0.946903 | ... | 0.948582 | 0.020478 | 1 | 0.993348 | 0.997783 | 0.995565 | 0.993348 | 0.995575 | 0.995124 | 0.001660 |
| 9 | 0.118126 | 0.017722 | 0.010340 | 0.001285 | 2 | 4 | 100 | {'min_samples_leaf': 2, 'min_samples_split': 4... | 0.938053 | 0.946903 | ... | 0.948582 | 0.020478 | 1 | 0.993348 | 0.997783 | 0.995565 | 0.993348 | 0.995575 | 0.995124 | 0.001660 |
| 11 | 0.044302 | 0.000977 | 0.004670 | 0.000183 | 2 | 6 | 40 | {'min_samples_leaf': 2, 'min_samples_split': 6... | 0.946903 | 0.946903 | ... | 0.948582 | 0.022660 | 1 | 0.991131 | 0.988914 | 0.993348 | 0.986696 | 0.988938 | 0.989805 | 0.002259 |
3 rows × 23 columns
#running random forest with optimized paramaters from grid search
forest2 = RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',
max_depth=None, max_features='auto', max_leaf_nodes=None,
min_impurity_decrease=0.0, min_impurity_split=None,
min_samples_leaf=2, min_samples_split=2,
min_weight_fraction_leaf=0.0, n_estimators=100,
n_jobs=None, oob_score=False, random_state=20, verbose=0,
warm_start=False)scoring = ['accuracy','precision', 'recall', 'f1']
rfc_cv=cross_validate(forest2, X_train,y_train, cv=5, scoring=scoring, return_train_score=True)type(rfc_cv)dict
rfc_cv_df=pd.DataFrame.from_dict(rfc_cv)
rfc_cv_df=rfc_cv_df.round(2)
rfc_cv_df.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| fit_time | score_time | test_accuracy | train_accuracy | test_precision | train_precision | test_recall | train_recall | test_f1 | train_f1 | |
|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0.13 | 0.04 | 0.94 | 0.99 | 0.91 | 0.99 | 0.96 | 1.0 | 0.94 | 0.99 |
| 1 | 0.11 | 0.04 | 0.95 | 1.00 | 0.95 | 1.00 | 0.95 | 1.0 | 0.95 | 1.00 |
| 2 | 0.12 | 0.04 | 0.92 | 1.00 | 0.86 | 0.99 | 1.00 | 1.0 | 0.92 | 1.00 |
| 3 | 0.11 | 0.04 | 0.98 | 0.99 | 0.98 | 0.99 | 0.98 | 1.0 | 0.98 | 0.99 |
| 4 | 0.15 | 0.04 | 0.96 | 1.00 | 0.93 | 0.99 | 0.98 | 1.0 | 0.95 | 1.00 |
rfc_cv_df.mean()fit_time 0.124
score_time 0.040
test_accuracy 0.950
train_accuracy 0.996
test_precision 0.926
train_precision 0.992
test_recall 0.974
train_recall 1.000
test_f1 0.948
train_f1 0.996
dtype: float64
##need to fix rounding of test scores
print('Average Train Scores: ', 'Accuracy: ', rfc_cv_df.train_accuracy.mean(), 'Precision: ', rfc_cv_df.train_precision.mean(), 'Recall: ', rfc_cv_df.train_recall.mean(), ' F1: ', rfc_cv_df.train_f1.mean())
print('Average Test Scores: ', 'Accuracy: ', rfc_cv_df.test_accuracy.mean(), 'Precision: ', rfc_cv_df.test_precision.mean(), 'Recall: ', rfc_cv_df.test_recall.mean(), 'F1: ', rfc_cv_df.test_f1.mean())Average Train Scores: Accuracy: 0.9960000000000001 Precision: 0.992 Recall: 1.0 F1: 0.9960000000000001
Average Test Scores: Accuracy: 0.95 Precision: 0.9259999999999999 Recall: 0.974 F1: 0.9480000000000001
## attempt to generate cross-validated prediction labels to generate a confusion matrix
rfc_cv_predict=cross_val_predict(X_train,y_train, cv=5)---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
<ipython-input-158-6b4265301f72> in <module>
1 ## attempt to generate cross-validated prediction labels to generate a confusion matrix
----> 2 rfc_cv_predict=cross_val_predict(X_train,y_train, cv=5)
~/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_validation.py in cross_val_predict(estimator, X, y, groups, cv, n_jobs, verbose, fit_params, pre_dispatch, method)
778 prediction_blocks = parallel(delayed(_fit_and_predict)(
779 clone(estimator), X, y, train, test, verbose, fit_params, method)
--> 780 for train, test in cv.split(X, y, groups))
781
782 # Concatenate the predictions
~/anaconda3/lib/python3.7/site-packages/joblib/parallel.py in __call__(self, iterable)
919 # remaining jobs.
920 self._iterating = False
--> 921 if self.dispatch_one_batch(iterator):
922 self._iterating = self._original_iterator is not None
923
~/anaconda3/lib/python3.7/site-packages/joblib/parallel.py in dispatch_one_batch(self, iterator)
752 tasks = BatchedCalls(itertools.islice(iterator, batch_size),
753 self._backend.get_nested_backend(),
--> 754 self._pickle_cache)
755 if len(tasks) == 0:
756 # No more tasks available in the iterator: tell caller to stop.
~/anaconda3/lib/python3.7/site-packages/joblib/parallel.py in __init__(self, iterator_slice, backend_and_jobs, pickle_cache)
208
209 def __init__(self, iterator_slice, backend_and_jobs, pickle_cache=None):
--> 210 self.items = list(iterator_slice)
211 self._size = len(self.items)
212 if isinstance(backend_and_jobs, tuple):
~/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_validation.py in <genexpr>(.0)
778 prediction_blocks = parallel(delayed(_fit_and_predict)(
779 clone(estimator), X, y, train, test, verbose, fit_params, method)
--> 780 for train, test in cv.split(X, y, groups))
781
782 # Concatenate the predictions
~/anaconda3/lib/python3.7/site-packages/sklearn/base.py in clone(estimator, safe)
58 "it does not seem to be a scikit-learn estimator "
59 "as it does not implement a 'get_params' methods."
---> 60 % (repr(estimator), type(estimator)))
61 klass = estimator.__class__
62 new_object_params = estimator.get_params(deep=False)
TypeError: Cannot clone object ' infant deaths Alcohol percentage expenditure Hepatitis B Measles \
2502 3 5.52 131.042127 88.0 37
1898 521 8.90 133.123087 49.0 8491
350 2 5.48 306.952735 87.0 1
229 0 14.44 8.494095 96.0 1
2739 5 7.99 29.381727 96.0 42724
2208 0 3.85 47.736382 46.0 0
1799 3 0.01 796.873426 84.0 86
338 2 0.01 1.117811 95.0 1
34 21 0.53 544.450743 95.0 25
1190 1100 3.00 64.605901 44.0 33634
553 2 7.33 1275.689625 95.0 0
1370 66 1.97 59.833614 81.0 1516
283 25 1.15 10.736281 75.0 262
2491 2 0.01 708.955665 98.0 0
2428 1 9.35 4255.781693 96.0 1204
1556 29 0.87 79.508825 74.0 6
1525 0 15.14 1807.071336 93.0 0
1668 0 0.01 115.278428 97.0 0
91 11 7.63 719.366380 81.0 0
1816 20 0.26 80.587884 9.0 3362
2155 14 8.34 9.074569 97.0 31
2848 0 0.83 361.094098 62.0 0
2365 0 0.99 229.668749 99.0 0
2373 0 1.18 16.831706 79.0 0
27 1 4.54 221.842800 99.0 7
637 1 4.04 1070.268999 86.0 1
235 1 12.05 42.334439 99.0 2
2133 4 9.78 180.109513 98.0 10
1982 11 0.81 103.727773 59.0 0
396 1 11.19 32.386161 96.0 0
... ... ... ... ... ...
1331 4 0.41 63.878452 98.0 20
23 1 5.61 36.622068 99.0 0
1622 54 0.61 101.811400 66.0 24
390 1 10.93 661.514433 96.0 2249
233 1 12.60 364.426052 98.0 149
1798 3 0.01 760.655055 89.0 1028
1899 527 9.05 14.567647 63.0 1272
638 1 4.17 112.949375 94.0 0
2153 12 0.01 11.710907 98.0 17
2053 3 7.74 466.738311 95.0 133
1627 57 0.53 67.709207 9.0 128
2438 2 12.26 228.354302 82.0 67
1822 31 0.20 45.879899 69.0 2838
857 6 0.49 11.765723 94.0 0
1784 47 0.33 21.236988 4.0 2046
1469 1 2.10 618.361486 75.0 213
859 7 0.97 10.602698 94.0 128
1033 0 9.51 3682.887170 95.0 1
566 248 4.88 50.283489 99.0 52461
1038 0 9.46 2124.921517 92.0 0
1619 52 0.01 46.562317 73.0 290
1943 359 0.01 62.293611 72.0 1370
2156 16 7.92 63.787236 97.0 121
361 70 7.10 30.303747 99.0 57
393 1 10.39 508.630459 96.0 1
2398 43 7.28 1038.885632 71.0 12499
2703 6 2.35 37.884661 97.0 0
138 0 12.40 5992.588029 86.0 9
683 0 11.41 1562.520827 88.0 1
83 9 8.35 1133.558003 91.0 2
BMI under-five deaths Polio Total expenditure Diphtheria \
2502 26.7 4 87.0 5.16 85.0
1898 22.2 817 54.0 3.47 54.0
350 3.5 4 97.0 5.73 97.0
229 59.3 1 99.0 5.55 98.0
2739 56.8 6 99.0 6.39 98.0
2208 7.9 0 49.0 5.25 53.0
1799 33.3 4 84.0 8.24 84.0
338 36.8 2 96.0 5.84 95.0
34 57.2 24 95.0 7.12 95.0
1190 16.4 1500 79.0 4.33 82.0
553 58.1 2 94.0 6.18 94.0
1370 17.3 100 76.0 4.80 81.0
283 2.1 39 74.0 4.56 72.0
2491 31.2 3 98.0 9.66 98.0
2428 64.8 2 97.0 9.39 97.0
1556 19.5 40 73.0 4.15 74.0
1525 6.9 0 93.0 6.67 93.0
1668 32.8 0 98.0 4.81 97.0
91 56.3 12 91.0 6.84 98.0
1816 17.4 25 9.0 5.89 9.0
2155 18.9 20 93.0 7.71 97.0
2848 48.2 0 66.0 3.90 66.0
2365 47.2 0 99.0 5.80 99.0
2373 4.4 0 84.0 6.00 84.0
27 48.9 1 98.0 6.38 97.0
637 48.3 1 88.0 8.45 88.0
235 56.2 1 99.0 6.59 99.0
2133 51.9 5 99.0 4.36 99.0
1982 43.4 14 76.0 4.60 63.0
396 58.6 1 96.0 7.43 96.0
... ... ... ... ... ...
1331 64.8 4 98.0 7.45 98.0
23 52.6 1 99.0 5.87 99.0
1622 21.3 90 72.0 6.59 66.0
390 62.1 1 94.0 6.78 94.0
233 57.2 1 97.0 6.34 99.0
1798 34.1 3 89.0 8.53 89.0
1899 21.6 832 66.0 4.24 63.0
638 47.3 1 94.0 8.23 94.0
2153 2.1 17 98.0 7.69 98.0
2053 53.6 3 98.0 5.86 98.0
1627 18.5 98 79.0 6.56 78.0
2438 58.8 2 98.0 7.25 98.0
1822 14.4 41 91.0 5.70 89.0
857 15.1 9 94.0 3.69 94.0
1784 2.5 61 9.0 1.87 84.0
1469 6.2 1 74.0 8.91 75.0
859 14.3 9 94.0 3.30 94.0
1033 62.4 0 99.0 9.76 99.0
566 27.3 288 99.0 5.80 99.0
1038 59.2 1 93.0 8.61 94.0
1619 23.2 85 74.0 6.86 73.0
1943 24.7 442 72.0 2.61 72.0
2156 18.3 23 93.0 7.91 97.0
361 48.6 79 99.0 8.36 99.0
393 6.3 1 96.0 6.67 95.0
2398 47.2 62 72.0 8.50 72.0
2703 43.4 8 97.0 1.88 96.0
138 52.7 0 86.0 1.53 86.0
683 55.3 0 98.0 6.37 98.0
83 61.0 10 99.0 5.20 91.0
HIV/AIDS GDP Population thinness 1-19 years \
2502 49.9 1324.996228 1.893000e+03 8.6
1898 4.8 2327.326700 1.585783e+08 11.3
350 37.2 3128.977930 1.754935e+06 11.8
229 0.1 63.388770 9.495830e+05 2.0
2739 0.8 233.188310 4.678775e+06 2.6
2208 0.1 322.543121 1.822860e+05 0.2
1799 3.7 5749.447520 2.263934e+06 9.5
338 2.8 77.625783 2.128570e+05 7.0
34 0.1 5471.866766 3.833856e+07 5.9
1190 0.2 1461.671957 1.247236e+08 26.9
553 0.1 9484.681227 1.631979e+07 0.9
1370 9.1 839.181117 3.885990e+05 8.4
283 2.1 583.493514 7.754000e+03 8.9
2491 9.8 3598.759720 1.271456e+06 4.5
2428 0.1 28562.293240 4.677355e+06 0.6
1556 0.4 461.723722 2.296115e+07 7.3
1525 0.1 14341.836000 2.987773e+06 2.7
1668 0.1 1153.938220 1.269340e+05 7.0
91 0.1 4251.574348 3.872870e+07 1.1
1816 0.2 681.792587 2.764992e+07 16.3
2155 1.3 617.317648 1.516710e+05 6.3
2848 0.1 2643.441423 2.378500e+04 1.5
2365 0.1 1642.837974 5.396140e+05 1.2
2373 0.1 744.765742 4.467690e+05 1.3
27 0.1 2416.588235 3.269390e+05 1.8
637 0.1 4167.714170 4.125971e+06 2.2
235 0.1 2378.339270 9.731460e+05 2.4
2133 0.1 1839.729450 2.213197e+06 3.8
1982 1.3 1178.724690 6.787187e+06 1.4
396 0.1 271.468240 7.775327e+06 2.3
... ... ... ... ...
1331 0.1 466.947750 8.893600e+04 3.9
23 0.1 437.539647 2.947314e+06 1.6
1622 1.5 835.889980 1.554989e+06 8.5
390 0.1 6955.987733 7.444443e+06 2.0
233 0.1 3848.215966 9.649240e+05 2.2
1798 2.5 5488.131712 2.316520e+05 9.0
1899 4.9 197.661422 1.544218e+07 11.7
638 0.1 462.149650 4.632400e+04 2.2
2153 0.5 688.876856 1.165151e+06 5.9
2053 0.1 4981.198619 3.824876e+06 2.5
1627 1.8 521.642580 1.322764e+06 9.7
2438 0.1 1719.535410 4.143156e+07 0.6
1822 0.2 348.631453 2.594618e+06 17.4
857 1.1 326.825642 4.232636e+06 9.1
1784 0.5 1186.423937 5.553310e+05 13.0
1469 0.1 5424.223560 3.863267e+06 4.7
859 1.4 297.828588 4.666480e+05 9.3
1033 0.1 31997.282100 1.177841e+06 0.8
566 0.1 3838.434292 1.331260e+05 4.4
1038 0.1 18477.578410 1.928700e+04 0.8
1619 1.6 825.572992 1.696285e+07 7.9
1943 0.1 1316.989660 1.855463e+08 19.4
2156 2.3 563.491487 1.246842e+06 6.5
361 0.1 586.145975 1.891241e+07 3.1
393 0.1 4513.136280 7.612200e+04 2.2
2398 11.0 7362.761390 5.979432e+06 7.3
2703 0.1 436.459223 5.795000e+03 3.2
138 0.1 38242.425200 8.227829e+06 1.7
683 0.1 25324.486660 1.276580e+05 0.9
83 0.1 12969.771200 4.296739e+06 1.0
thinness 5-9 years Schooling Developed Developing
2502 8.8 9.2 0 1
1898 11.2 9.5 0 1
350 11.8 11.8 0 1
229 2.2 15.5 0 1
2739 2.7 14.7 0 1
2208 0.2 12.9 0 1
1799 9.4 11.5 0 1
338 6.7 12.6 0 1
34 5.8 14.4 0 1
1190 27.7 10.8 0 1
553 0.9 14.9 0 1
1370 8.3 10.1 0 1
283 8.8 8.1 0 1
2491 4.6 11.4 0 1
2428 0.5 17.2 1 0
1556 7.2 10.3 0 1
1525 2.7 16.5 1 0
1668 6.9 14.7 0 1
91 1.0 16.3 0 1
1816 16.7 12.3 0 1
2155 6.2 10.2 0 1
2848 1.5 10.7 0 1
2365 1.2 9.4 0 1
2373 1.3 8.0 0 1
27 1.9 10.9 0 1
637 2.1 12.1 0 1
235 2.5 14.1 0 1
2133 4.2 11.7 1 0
1982 1.3 8.9 0 1
396 2.4 12.9 1 0
... ... ... ... ...
1331 3.9 13.1 0 1
23 1.6 12.0 0 1
1622 8.3 7.5 0 1
390 2.1 13.8 1 0
233 2.4 14.6 0 1
1798 8.9 11.6 0 1
1899 11.6 9.3 0 1
638 2.2 11.9 0 1
2153 5.9 10.8 0 1
2053 2.7 14.7 1 0
1627 9.5 6.1 0 1
2438 0.5 15.7 1 0
1822 18.0 9.6 0 1
857 9.1 5.2 0 1
1784 13.3 9.1 0 1
1469 4.6 14.2 0 1
859 9.3 5.3 0 1
1033 0.7 15.9 0 1
566 3.8 12.2 0 1
1038 0.8 15.2 0 1
1619 7.7 8.2 0 1
1943 19.8 7.8 0 1
2156 6.3 10.0 0 1
361 3.1 13.8 0 1
393 2.2 13.5 1 0
2398 8.9 12.8 0 1
2703 3.3 10.5 0 1
138 1.9 14.9 1 0
683 1.0 13.5 1 0
83 0.9 17.2 0 1
[564 rows x 18 columns]' (type <class 'pandas.core.frame.DataFrame'>): it does not seem to be a scikit-learn estimator as it does not implement a 'get_params' methods.
type(rfc_cv_predict)forest2.fit(X_train,y_train)RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',
max_depth=None, max_features='auto', max_leaf_nodes=None,
min_impurity_decrease=0.0, min_impurity_split=None,
min_samples_leaf=2, min_samples_split=2,
min_weight_fraction_leaf=0.0, n_estimators=100,
n_jobs=None, oob_score=False, random_state=20, verbose=0,
warm_start=False)
y_pred_test=forest2.predict(X_test)forest2.score(X_test,y_test)0.91005291005291
acc = accuracy_score(y_pred_test,y_test) * 100
print("Accuracy is :{0}".format(acc))
# Check the AUC for predictions
false_positive_rate, true_positive_rate, thresholds = roc_curve(y_test, y_pred_test)
roc_auc = auc(false_positive_rate, true_positive_rate)
print("\nAUC is :{0}".format(round(roc_auc,2)))
# Create and print a confusion matrix
print('\nConfusion Matrix')
print('----------------')
pd.crosstab(y_test, y_pred_test, rownames=['True'], colnames=['Predicted'], margins=True)Accuracy is :91.005291005291
AUC is :0.92
Confusion Matrix
----------------
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Predicted | 0 | 1 | All |
|---|---|---|---|
| True | |||
| 0 | 85 | 16 | 101 |
| 1 | 1 | 87 | 88 |
| All | 86 | 103 | 189 |
def plot_feature_importances(model):
n_features = X_train.shape[1]
plt.figure(figsize=(8,8))
plt.barh(range(n_features), model.feature_importances_, align='center')
plt.yticks(np.arange(n_features), X_train.columns.values)
plt.xlabel("Feature importance")
plt.ylabel("Feature")
plot_feature_importances(forest2)
We want to see how the model performs when presented with data from countries it has never seen. If the performance doesn't drop much, the country has little effect on the life expectancy.
df_country_split_b.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Country | Year | target | Adult Mortality | infant deaths | Alcohol | percentage expenditure | Hepatitis B | Measles | BMI | ... | Diphtheria | HIV/AIDS | GDP | Population | thinness 1-19 years | thinness 5-9 years | Income composition of resources | Schooling | Developed | Developing | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Afghanistan | 2015 | 0 | 263.0 | 62 | 0.01 | 71.279624 | 65.0 | 1154 | 19.1 | ... | 65.0 | 0.1 | 584.259210 | 33736494.0 | 17.2 | 17.3 | 0.479 | 10.1 | 0 | 1 |
| 1 | Afghanistan | 2014 | 0 | 271.0 | 64 | 0.01 | 73.523582 | 62.0 | 492 | 18.6 | ... | 62.0 | 0.1 | 612.696514 | 327582.0 | 17.5 | 17.5 | 0.476 | 10.0 | 0 | 1 |
| 2 | Afghanistan | 2013 | 0 | 268.0 | 66 | 0.01 | 73.219243 | 64.0 | 430 | 18.1 | ... | 64.0 | 0.1 | 631.744976 | 31731688.0 | 17.7 | 17.7 | 0.470 | 9.9 | 0 | 1 |
| 3 | Afghanistan | 2012 | 0 | 272.0 | 69 | 0.01 | 78.184215 | 67.0 | 2787 | 17.6 | ... | 67.0 | 0.1 | 669.959000 | 3696958.0 | 17.9 | 18.0 | 0.463 | 9.8 | 0 | 1 |
| 4 | Afghanistan | 2011 | 0 | 275.0 | 71 | 0.01 | 7.097109 | 68.0 | 3013 | 17.2 | ... | 68.0 | 0.1 | 63.537231 | 2978599.0 | 18.2 | 18.2 | 0.454 | 9.5 | 0 | 1 |
5 rows × 23 columns
y_test_2 = df_country_split_a['target']
X_test_2 = df_country_split_a.drop(['Country', 'target','Adult Mortality','Income composition of resources', 'Year'], axis=1)
y_pred_2 = forest2.predict(X_test_2)
acc = accuracy_score(y_test_2,y_pred_2) * 100
print("Accuracy is :{0}".format(acc))
# Check the AUC for predictions
false_positive_rate, true_positive_rate, thresholds = roc_curve(y_test_2, y_pred_2)
roc_auc = auc(false_positive_rate, true_positive_rate)
print("\nAUC is :{0}".format(round(roc_auc,2)))
# Create and print a confusion matrix
print('\nConfusion Matrix')
print('----------------')
pd.crosstab(y_test_2, y_pred_2, rownames=['True'], colnames=['Predicted'], margins=True)Accuracy is :97.34395750332006
AUC is :0.97
Confusion Matrix
----------------
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| Predicted | 0 | 1 | All |
|---|---|---|---|
| True | |||
| 0 | 372 | 19 | 391 |
| 1 | 1 | 361 | 362 |
| All | 373 | 380 | 753 |
It's 88% accurate on countries it is never seen before when 70 countries (36%) were completely withheld from training.
When we increased the number to 100 countries (trained on the remaining), the accuracy dropped to 85%
Our model, in addition to life expectancy, is able to draw attention to some modifiable features/areas countries' can focus on regardless of their classification
Features of high importance in our classification model included schooling length (in years), vaccination rates (diphtheria, pertussis, tetanus, measles, mumps, rubella, polio), HIV/AID prevalence, and weight