HW5_visualization_and_eda

Homework Assignment #5 : Visualization and EDA

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

%matplotlib inline

From the given data (covid20200416_rev.csv), write your codes according to the following questions.

df_covid = pd.read_csv('./covid20200416_rev.csv')
print(len(df_covid))
df_covid.head()

` 212 `

	country	total_cases	total_deaths	total_recovered	Tot_cases_per_1Mpop	Deaths_per_1Mpop	region
0	USA	644089	28529	48701	1946.0	86.0	North America
1	Spain	180659	18812	70853	3864.0	402.0	Europe
2	Italy	165155	21645	38092	2732.0	358.0	Europe
3	France	147863	17167	30955	2265.0	263.0	Europe
4	Germany	134753	3804	72600	1608.0	45.0	Europe

df_covid.describe()

	total_cases	total_deaths	total_recovered	Tot_cases_per_1Mpop	Deaths_per_1Mpop
count	212.000000	212.000000	212.000000	210.000000	162.000000
mean	9826.905660	634.976415	2407.268868	626.120143	33.438457
std	50052.898899	3155.560189	10488.024252	1463.639958	107.465073
min	1.000000	0.000000	0.000000	0.030000	0.030000
25%	39.500000	1.000000	5.000000	18.000000	0.600000
50%	382.500000	6.000000	67.000000	104.500000	3.000000
75%	2270.250000	52.750000	346.250000	465.750000	17.000000
max	644089.000000	28529.000000	77892.000000	11582.000000	1061.000000

(1) Visualize the mean total_cases according to each region by the bar chart.

# your code here
mean_total_cases_by_region = df_covid.groupby('region').total_cases.mean()
mean_total_cases_by_region
mean_total_cases_by_region.sort_values(inplace=True, ascending=True)
x = mean_total_cases_by_region.index.tolist()
y = mean_total_cases_by_region.values.tolist()
# x1 = list(mean_total_cases_by_region.index)
# y1 = list(mean_total_cases_by_region.values)
#print(type(x), y, type(x1), y1)

plt.title('Mean Total Cases by Each Region')
plt.xlabel("Region")
plt.ylabel("Mean cases")
#plt.barh(x,y)
plt.bar(x,y)
plt.xticks(rotation=-15)
plt.show()

(2) Plot the distribution(histogram) of total cases per 1M population (Tot_cases_per_1Mpop) using Seaborn package.

• Note: Before applying the visualization for distribution, you may need to filter out rows of missing value (NaN).

# your code here 2-1 : find NaN
print("duplicated data?:", df_covid.duplicated().sum())
sns.heatmap(df_covid.isnull(), cbar=False)
plt.show()

` duplicated data?: 0 `

# your code here 2-2 :dropna 
df_covid = df_covid.dropna(subset=['Tot_cases_per_1Mpop'], axis = 'index')

d = df_covid.Tot_cases_per_1Mpop.isnull()
# len(d)
count = 0 ;
for i in range(0, len(d)):
    if d.iloc[i] == True:
        count += 1 
print("NaN of 'Tot_cases_per_1Mpop':", count)

sns.heatmap(df_covid.isnull(), cbar=False)
plt.show()

` NaN of ‘Tot_cases_per_1Mpop’: 0 `

# your code here 2-3 : draw plot
sns.distplot(df_covid['Tot_cases_per_1Mpop'])
plt.show()

(3) Plot the scatter plot according to the following conditions

• x values: ‘total_cases’
• y values: ‘total_deaths’

# your code here 3-1 : describe dataframe
df_covid.describe()

	total_cases	total_deaths	total_recovered	Tot_cases_per_1Mpop	Deaths_per_1Mpop
count	210.000000	210.000000	210.000000	210.000000	162.000000
mean	9917.061905	640.957143	2427.128571	626.120143	33.438457
std	50283.196920	3170.021744	10536.046314	1463.639958	107.465073
min	1.000000	0.000000	0.000000	0.030000	0.030000
25%	41.500000	1.000000	5.250000	18.000000	0.600000
50%	382.500000	6.000000	67.000000	104.500000	3.000000
75%	2426.750000	58.250000	333.000000	465.750000	17.000000
max	644089.000000	28529.000000	77892.000000	11582.000000	1061.000000

# your code here 3-2 : draw plot

x = df_covid['total_cases']
y = df_covid['total_deaths']

plt.scatter(x, y, marker = '+', s=150, color='red')
plt.show()

(4) Plot the heatmap of correlations among all numerical variables. Which variable (column) is the most correlated with the value “total_cases”?

# your code here : 4-1 calculate corr
df_corr = df_covid[:]
df_corr.corr()

	total_cases	total_deaths	total_recovered	Tot_cases_per_1Mpop	Deaths_per_1Mpop
total_cases	1.000000	0.885806	0.652631	0.156965	0.220542
total_deaths	0.885806	1.000000	0.664417	0.209220	0.379455
total_recovered	0.652631	0.664417	1.000000	0.160642	0.244011
Tot_cases_per_1Mpop	0.156965	0.209220	0.160642	1.000000	0.852964
Deaths_per_1Mpop	0.220542	0.379455	0.244011	0.852964	1.000000

# your code here :4-2 draw heatmap
h = sns.heatmap(df_corr.corr(), annot=True, cmap='RdYlGn_r')
h.set_xticklabels(h.get_xticklabels(),rotation=-15, fontsize='small')
plt.show()

5) Plot the boxplots of mean total_cases according to each “region”.

# your code here
sns.catplot(x='region', y='total_cases', kind='box', data=df_covid)
plt.xticks(rotation = -15)
plt.show()

Open question: Can you find any other insight from the data?

#your code here : Open question-1
sns.pairplot(data=df_covid, hue='region')
plt.show()

#your code here : Open question-2
df_covid.sort_values(by='total_deaths', ascending=False)

	country	total_cases	total_deaths	total_recovered	Tot_cases_per_1Mpop	Deaths_per_1Mpop	region
0	USA	644089	28529	48701	1946.00	86.0	North America
2	Italy	165155	21645	38092	2732.00	358.0	Europe
1	Spain	180659	18812	70853	3864.00	402.0	Europe
3	France	147863	17167	30955	2265.00	263.0	Europe
5	UK	98476	12868	344	1451.00	190.0	Europe
...	...	...	...	...	...	...	...
175	Nepal	16	0	1	0.50	NaN	Asia
176	Dominica	16	0	8	222.00	NaN	North America
178	Namibia	16	0	3	6.00	NaN	Africa
179	Saint Lucia	15	0	11	82.00	NaN	North America
211	Yemen	1	0	0	0.03	NaN	Asia

210 rows × 7 columns

#your code here : Open question-3
df_covid.sort_values(by='total_recovered', ascending=False)

	country	total_cases	total_deaths	total_recovered	Tot_cases_per_1Mpop	Deaths_per_1Mpop	region
6	China	82341	3342	77892	57.00	2.0	Asia
4	Germany	134753	3804	72600	1608.00	45.0	Europe
1	Spain	180659	18812	70853	3864.00	402.0	Europe
7	Iran	76389	4777	49933	909.00	57.0	Asia
0	USA	644089	28529	48701	1946.00	86.0	North America
...	...	...	...	...	...	...	...
172	Belize	18	2	0	45.00	5.0	North America
158	Haiti	41	3	0	4.00	0.3	North America
157	Guinea-Bissau	43	0	0	22.00	NaN	Africa
149	French Polynesia	55	0	0	196.00	NaN	Oceania
211	Yemen	1	0	0	0.03	NaN	Asia

210 rows × 7 columns

#your code here : Open question-4
df_covid_no_usa = df_covid.drop(df_covid[df_covid['country'] == 'USA'].index)
sns.pairplot(data=df_covid_no_usa, hue='region')
plt.show()

`

결론

• North America(미국) 이 다른 나라에 비해 월등하게 높은 감염이 일어나고 있다.
• 미국 케이스를 제외하면 total cases 기준 유럽국가들의 사망률(특히 이탈리아)이 높다.
• 아시아 국가들(중국, 이란)이 동일 감염case에 대해 recover 되는 확률이 높은 편이다 `

---