data science python

肺炎疫情 (COVID-19) 到 2020 年 12 月的現狀 (自 Kaggle Notebook 翻譯)

Posted by blueskyson on January 29, 2021

來源: https://www.kaggle.com/corochann/covid-19-current-situation-on-december/notebook
此筆記的資料集來源: Johns Hopkins University github repository
最後更新: 2021-02-01

主要重點

  1. 綜觀全球疫情走勢
    • 以 折線圖 分析
  2. 分析各國確診與死亡人數變化
    • 以 柱狀圖、顏色指標全球動態地圖、顏色指標折線圖、顏色指標熱圖 分析
  3. 各國省分詳細資訊
    • 有多少國家提供精確的省分疫情資料
    • 以 Dictionary 分析
  4. 深入探討美國疫情
    • 以 表格、顏色指標美國動態地圖、顏色指標折線圖、顏色指標熱圖 分析
  5. 分析歐洲疫情
    • 以 顏色指標歐洲動態地圖、顏色指標折線圖、顏色指標熱圖 分析
  6. 分析亞洲疫情
    • 以 顏色指標亞洲動態地圖、顏色指標折線圖、顏色指標熱圖 分析
  7. 哪些國家已經脫離疫情高峰
    • 以 顏色指標全球動態地圖 分析
  8. 全球疫情何時趨緩
    • 使用 sigmoid 函數預測
    • 以 顏色指標折線圖 分析

安裝

pip install -r requirements.txt

In [1]:
import gc
import os
from pathlib import Path
import random
import sys

from tqdm import tqdm
import numpy as np
import pandas as pd
import scipy as sp


import matplotlib.pyplot as plt
import seaborn as sns

from IPython.core.display import display, HTML

# --- plotly ---
from plotly import tools, subplots
import plotly.offline as py
py.init_notebook_mode(connected=True)
import plotly.graph_objs as go
import plotly.express as px
import plotly.figure_factory as ff
import plotly.io as pio
pio.templates.default = "plotly_dark"

# --- models ---
from sklearn import preprocessing
from sklearn.model_selection import KFold
import lightgbm as lgb
import xgboost as xgb
import catboost as cb

# --- setup ---
pd.set_option('max_columns', 50)

載入資料

使用到的資料集: COVID-19/csse_covid_19_data/csse_covid_19_time_series/

  • 全球

    • confirmed_global_df : dataframe，儲存從 2020/1/22 到現在的各國確診人數

    • deaths_global_df : dataframe，儲存從 2020/1/22 到現在各國死亡人數

    • recovered_global_df : dataframe，儲存從 2020/1/22 到現在各國康復人數

  • 美國

    • confirmed_us_df : dataframe，儲存從 2020/1/22 到現在美國確診人數

    • deaths_us_df : dataframe，儲存從 2020/1/22 到現在美國死亡人數

In [2]:
import requests

for filename in ['time_series_covid19_confirmed_global.csv',
               'time_series_covid19_deaths_global.csv',
               'time_series_covid19_recovered_global.csv',
               'time_series_covid19_confirmed_US.csv',
               'time_series_covid19_deaths_US.csv']:
    print(f'Downloading {filename}')
    url = f'https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/{filename}'
    myfile = requests.get(url)
    open(filename, 'wb').write(myfile.content)

Downloading time_series_covid19_confirmed_global.csv
Downloading time_series_covid19_deaths_global.csv
Downloading time_series_covid19_recovered_global.csv
Downloading time_series_covid19_confirmed_US.csv
Downloading time_series_covid19_deaths_US.csv

資料前處理

變更日期的格式，由 mm/dd/yy 改成 yy-mm-dd

In [3]:
from datetime import datetime

def _convert_date_str(df):
  try:
    df.columns = list(df.columns[:4]) + [datetime.strptime(d, "%m/%d/%y").date().strftime("%Y-%m-%d") for d in df.columns[4:]]
  except:
    print('_convert_date_str failed with %y, try %Y')
  df.columns = list(df.columns[:4]) + [datetime.strptime(d, "%m/%d/%Y").date().strftime("%Y-%m-%d") for d in df.columns[4:]]


confirmed_global_df = pd.read_csv('time_series_covid19_confirmed_global.csv')
_convert_date_str(confirmed_global_df)

deaths_global_df = pd.read_csv('time_series_covid19_deaths_global.csv')
_convert_date_str(deaths_global_df)

recovered_global_df = pd.read_csv('time_series_covid19_recovered_global.csv')
_convert_date_str(recovered_global_df)

將所有日期合併到同一欄位 (Date) ，該日期的累積人數合併到另一欄位 (Confirmed, Deaths, Recovered)

In [4]:
# Filter out problematic data points (The West Bank and Gaza had a negative value, cruise ships were associated with Canada, etc.)
removed_states = "Recovered|Grand Princess|Diamond Princess"
removed_countries = "US|The West Bank and Gaza"

confirmed_global_df.rename(columns={"Province/State": "Province_State", "Country/Region": "Country_Region"}, inplace=True)
deaths_global_df.rename(columns={"Province/State": "Province_State", "Country/Region": "Country_Region"}, inplace=True)
recovered_global_df.rename(columns={"Province/State": "Province_State", "Country/Region": "Country_Region"}, inplace=True)

confirmed_global_df = confirmed_global_df[~confirmed_global_df["Province_State"].replace(np.nan, "nan").str.match(removed_states)]
deaths_global_df    = deaths_global_df[~deaths_global_df["Province_State"].replace(np.nan, "nan").str.match(removed_states)]
recovered_global_df = recovered_global_df[~recovered_global_df["Province_State"].replace(np.nan, "nan").str.match(removed_states)]

confirmed_global_df = confirmed_global_df[~confirmed_global_df["Country_Region"].replace(np.nan, "nan").str.match(removed_countries)]
deaths_global_df    = deaths_global_df[~deaths_global_df["Country_Region"].replace(np.nan, "nan").str.match(removed_countries)]
recovered_global_df = recovered_global_df[~recovered_global_df["Country_Region"].replace(np.nan, "nan").str.match(removed_countries)]

將所有日期合併到同一欄位 (Date) ，該日期的累積人數合併到另一欄位 (Confirmed, Deaths, Recovered)

In [5]:
confirmed_global_melt_df = confirmed_global_df.melt(
    id_vars=['Country_Region', 'Province_State', 'Lat', 'Long'], value_vars=confirmed_global_df.columns[4:], var_name='Date', value_name='ConfirmedCases')

deaths_global_melt_df = deaths_global_df.melt(
    id_vars=['Country_Region', 'Province_State', 'Lat', 'Long'], value_vars=confirmed_global_df.columns[4:], var_name='Date', value_name='Deaths')

recovered_global_melt_df = deaths_global_df.melt(
    id_vars=['Country_Region', 'Province_State', 'Lat', 'Long'], value_vars=confirmed_global_df.columns[4:], var_name='Date', value_name='Recovered')

recovered_global_melt_df.head()
Out[5]:
Country_RegionProvince_StateLatLongDateRecovered
0AfghanistanNaN33.000065.00002020-01-220
1AlbaniaNaN41.153320.16832020-01-220
2AlgeriaNaN28.03391.65962020-01-220
3AndorraNaN42.50631.52182020-01-220
4AngolaNaN-11.202717.87392020-01-220

製作訓練用的美國的資料集

In [6]:
train = confirmed_global_melt_df.merge(deaths_global_melt_df, on=['Country_Region', 'Province_State', 'Lat', 'Long', 'Date'])
train = train.merge(recovered_global_melt_df, on=['Country_Region', 'Province_State', 'Lat', 'Long', 'Date'])
In [7]:
# --- US ---
confirmed_us_df = pd.read_csv('time_series_covid19_confirmed_US.csv')
deaths_us_df = pd.read_csv('time_series_covid19_deaths_US.csv')

# 丟掉不須用到的欄位
confirmed_us_df.drop(['UID', 'iso2', 'iso3', 'code3', 'FIPS', 'Admin2', 'Combined_Key'], inplace=True, axis=1)
deaths_us_df.drop(['UID', 'iso2', 'iso3', 'code3', 'FIPS', 'Admin2', 'Combined_Key', 'Population'], inplace=True, axis=1)

# 將 Long_ 欄位改名為 Long
confirmed_us_df.rename({'Long_': 'Long'}, axis=1, inplace=True)
deaths_us_df.rename({'Long_': 'Long'}, axis=1, inplace=True)

# 變更日期的格式，由 mm/dd/yy 改成 yy-mm-dd
_convert_date_str(confirmed_us_df)
_convert_date_str(deaths_us_df)

# clean
# 丟掉不屬於美國的地區
confirmed_us_df = confirmed_us_df[~confirmed_us_df.Province_State.str.match("Diamond Princess|Grand Princess|Recovered|Northern Mariana Islands|American Samoa")]
deaths_us_df = deaths_us_df[~deaths_us_df.Province_State.str.match("Diamond Princess|Grand Princess|Recovered|Northern Mariana Islands|American Samoa")]

# --- Aggregate by province state ---
#confirmed_us_df.groupby(['Country_Region', 'Province_State'])
confirmed_us_df = confirmed_us_df.groupby(['Country_Region', 'Province_State']).sum().reset_index()
deaths_us_df = deaths_us_df.groupby(['Country_Region', 'Province_State']).sum().reset_index()

# remove lat, long.
confirmed_us_df.drop(['Lat', 'Long'], inplace=True, axis=1)
deaths_us_df.drop(['Lat', 'Long'], inplace=True, axis=1)

# 合併日期
confirmed_us_melt_df = confirmed_us_df.melt(
    id_vars=['Country_Region', 'Province_State'], value_vars=confirmed_us_df.columns[2:], var_name='Date', value_name='ConfirmedCases')
deaths_us_melt_df = deaths_us_df.melt(
    id_vars=['Country_Region', 'Province_State'], value_vars=deaths_us_df.columns[2:], var_name='Date', value_name='Deaths')

# 將美國確診及死亡的資料合併
train_us = confirmed_us_melt_df.merge(deaths_us_melt_df, on=['Country_Region', 'Province_State', 'Date'])
In [8]:
train = pd.concat([train, train_us], axis=0, sort=False)

train_us.rename({'Country_Region': 'country', 'Province_State': 'province', 'Date': 'date', 'ConfirmedCases': 'confirmed', 'Deaths': 'fatalities'}, axis=1, inplace=True)
train_us['country_province'] = train_us['country'].fillna('') + '/' + train_us['province'].fillna('')
In [9]:
train
Out[9]:
Country_RegionProvince_StateLatLongDateConfirmedCasesDeathsRecovered
0AfghanistanNaN33.000065.00002020-01-22000.0
1AlbaniaNaN41.153320.16832020-01-22000.0
2AlgeriaNaN28.03391.65962020-01-22000.0
3AndorraNaN42.50631.52182020-01-22000.0
4AngolaNaN-11.202717.87392020-01-22000.0
...........................
7825USVirginiaNaNNaN2020-06-14545061546NaN
7826USWashingtonNaNNaN2020-06-14258341217NaN
7827USWest VirginiaNaNNaN2020-06-14229088NaN
7828USWisconsinNaNNaN2020-06-1422758692NaN
7829USWyomingNaNNaN2020-06-14106018NaN

45965 rows × 8 columns

In [10]:
train.rename({'Country_Region': 'country', 'Province_State': 'province', 'Id': 'id', 'Date': 'date', 'ConfirmedCases': 'confirmed', 'Deaths': 'fatalities', 'Recovered': 'recovered'}, axis=1, inplace=True)
train['country_province'] = train['country'].fillna('') + '/' + train['province'].fillna('')

# test.rename({'Country_Region': 'country', 'Province_State': 'province', 'Id': 'id', 'Date': 'date', 'ConfirmedCases': 'confirmed', 'Fatalities': 'fatalities'}, axis=1, inplace=True)
# test['country_province'] = test['country'].fillna('') + '/' + test['province'].fillna('')

綜觀全球疫情走勢

觀察 ww_df 的各個屬性

In [11]:
ww_df = train.groupby('date')[['confirmed', 'fatalities']].sum().reset_index()
ww_df['new_case'] = ww_df['confirmed'] - ww_df['confirmed'].shift(1)
ww_df['growth_factor'] = ww_df['new_case'] / ww_df['new_case'].shift(1)
ww_df.tail()
Out[11]:
dateconfirmedfatalitiesnew_casegrowth_factor
1402020-06-107376138417060133641.01.083561
1412020-06-117514529421455138391.01.035543
1422020-06-127632607425388118078.00.853220
1432020-06-137766757429730134150.01.136113
1442020-06-147900729433060133972.00.998673

把 ww_df 的 confirmed fatalities new_case 三個屬性分別出來，放在 ww_melt_df

In [12]:
ww_melt_df = pd.melt(ww_df, id_vars=['date'], value_vars=['confirmed', 'fatalities', 'new_case'])
ww_melt_df
Out[12]:
datevariablevalue
02020-01-22confirmed555.0
12020-01-23confirmed654.0
22020-01-24confirmed941.0
32020-01-25confirmed1434.0
42020-01-26confirmed2118.0
............
4302020-06-10new_case133641.0
4312020-06-11new_case138391.0
4322020-06-12new_case118078.0
4332020-06-13new_case134150.0
4342020-06-14new_case133972.0

435 rows × 3 columns

全球確診/死亡案例 (折線圖)

  • 2020/04/02 確診人數突破 1M，且死亡人數為 52K
  • 2020/07/28 確診人數突破 10M，且死亡人數為 500K
  • 2020/08/13 確診人數突破 20M，且死亡人數為 750K
  • 2020/11/08 確診人數突破 50M，且死亡人數為 1.25M
In [13]:
fig = px.line(ww_melt_df, x="date", y="value", color='variable', 
              title="Worldwide Confirmed/Death Cases Over Time")
fig.show()

全球確診/死亡案例 (折線圖) (取對數)

如果某段期間為斜直線，代表該段期間的案例為指數增長
比較 2020/3 初和 2020/3 底，確診案例成長率的上升速度略為增加

In [14]:
fig = px.line(ww_melt_df, x="date", y="value", color='variable',
              title="Worldwide Confirmed/Death Cases Over Time (Log scale)",log_y=True)
fig.show()

全球死亡率 (折線圖)

可以明顯看到，死亡率在 2020 年 5 月達到高峰 (7%)，隨後開始下降，可能是因為醫療能量慢慢提升，或是更多潛在患者確診?

In [15]:
ww_df['mortality'] = ww_df['fatalities'] / ww_df['confirmed']
fig = px.line(ww_df, x="date", y="mortality", title="Worldwide Mortality Rate Over Time")
fig.show()

計算全球新增確診案例成長因子 (折線圖)

公式: 每日新增確診案例 / 昨日新增確診案例

  • 大於1: 確診案例增加
  • 小於1: 確診案例減少
In [16]:
ww_df['mortality'] = ww_df['fatalities'] / ww_df['confirmed']
fig = px.line(ww_df, x="date", y="mortality", title="Worldwide Mortality Rate Over Time")
fig.show()

分析各國確診與死亡人數變化

宣告 country_df 條列各國確診與死亡人數，並查看 country_df 的各個欄位

In [17]:
country_df = train.groupby(['date', 'country'])[['confirmed', 'fatalities']].sum().reset_index()
country_df.tail()
Out[17]:
date country confirmed fatalities
72187 2021-01-31 Vietnam 1817 35
72188 2021-01-31 West Bank and Gaza 158962 1833
72189 2021-01-31 Yemen 2121 615
72190 2021-01-31 Zambia 54217 763
72191 2021-01-31 Zimbabwe 33388 1217

列出 dataset 的所有國家 (注意，台灣的標籤為 Taiwan* )

In [18]:
countries = country_df['country'].unique()
print(f'{len(countries)} countries are in dataset:\n{countries}')
192 countries are in dataset:
['Afghanistan' 'Albania' 'Algeria' 'Andorra' 'Angola'
 'Antigua and Barbuda' 'Argentina' 'Armenia' 'Australia' 'Austria'
 'Azerbaijan' 'Bahamas' 'Bahrain' 'Bangladesh' 'Barbados' 'Belarus'
 'Belgium' 'Belize' 'Benin' 'Bhutan' 'Bolivia' 'Bosnia and Herzegovina'
 'Botswana' 'Brazil' 'Brunei' 'Bulgaria' 'Burkina Faso' 'Burma' 'Burundi'
 'Cabo Verde' 'Cambodia' 'Cameroon' 'Canada' 'Central African Republic'
 'Chad' 'Chile' 'China' 'Colombia' 'Comoros' 'Congo (Brazzaville)'
 'Congo (Kinshasa)' 'Costa Rica' "Cote d'Ivoire" 'Croatia' 'Cuba' 'Cyprus'
 'Czechia' 'Denmark' 'Diamond Princess' 'Djibouti' 'Dominica'
 'Dominican Republic' 'Ecuador' 'Egypt' 'El Salvador' 'Equatorial Guinea'
 'Eritrea' 'Estonia' 'Eswatini' 'Ethiopia' 'Fiji' 'Finland' 'France'
 'Gabon' 'Gambia' 'Georgia' 'Germany' 'Ghana' 'Greece' 'Grenada'
 'Guatemala' 'Guinea' 'Guinea-Bissau' 'Guyana' 'Haiti' 'Holy See'
 'Honduras' 'Hungary' 'Iceland' 'India' 'Indonesia' 'Iran' 'Iraq'
 'Ireland' 'Israel' 'Italy' 'Jamaica' 'Japan' 'Jordan' 'Kazakhstan'
 'Kenya' 'Korea, South' 'Kosovo' 'Kuwait' 'Kyrgyzstan' 'Laos' 'Latvia'
 'Lebanon' 'Lesotho' 'Liberia' 'Libya' 'Liechtenstein' 'Lithuania'
 'Luxembourg' 'MS Zaandam' 'Madagascar' 'Malawi' 'Malaysia' 'Maldives'
 'Mali' 'Malta' 'Marshall Islands' 'Mauritania' 'Mauritius' 'Mexico'
 'Micronesia' 'Moldova' 'Monaco' 'Mongolia' 'Montenegro' 'Morocco'
 'Mozambique' 'Namibia' 'Nepal' 'Netherlands' 'New Zealand' 'Nicaragua'
 'Niger' 'Nigeria' 'North Macedonia' 'Norway' 'Oman' 'Pakistan' 'Panama'
 'Papua New Guinea' 'Paraguay' 'Peru' 'Philippines' 'Poland' 'Portugal'
 'Qatar' 'Romania' 'Russia' 'Rwanda' 'Saint Kitts and Nevis' 'Saint Lucia'
 'Saint Vincent and the Grenadines' 'Samoa' 'San Marino'
 'Sao Tome and Principe' 'Saudi Arabia' 'Senegal' 'Serbia' 'Seychelles'
 'Sierra Leone' 'Singapore' 'Slovakia' 'Slovenia' 'Solomon Islands'
 'Somalia' 'South Africa' 'South Sudan' 'Spain' 'Sri Lanka' 'Sudan'
 'Suriname' 'Sweden' 'Switzerland' 'Syria' 'Taiwan*' 'Tajikistan'
 'Tanzania' 'Thailand' 'Timor-Leste' 'Togo' 'Trinidad and Tobago'
 'Tunisia' 'Turkey' 'US' 'Uganda' 'Ukraine' 'United Arab Emirates'
 'United Kingdom' 'Uruguay' 'Uzbekistan' 'Vanuatu' 'Venezuela' 'Vietnam'
 'West Bank and Gaza' 'Yemen' 'Zambia' 'Zimbabwe']

列出所有國家確診案例的數量級

In [19]:
target_date = country_df['date'].max()
        
print('Date: ', target_date)
for i in [1, 10, 100, 1000, 10000, 100000, 1000000, 10000000]:
    n_countries = len(country_df.query('(date == @target_date) & confirmed > @i'))
    print(f'{n_countries} countries have more than {i} confirmed cases')
Date:  2021-01-31
190 countries have more than 1 confirmed cases
187 countries have more than 10 confirmed cases
181 countries have more than 100 confirmed cases
169 countries have more than 1000 confirmed cases
132 countries have more than 10000 confirmed cases
82 countries have more than 100000 confirmed cases
19 countries have more than 1000000 confirmed cases
2 countries have more than 10000000 confirmed cases

以柱狀圖表示

In [20]:
sns.set_style('white')
ax = sns.distplot(np.log10(country_df.query('date == @target_date')['confirmed'] + 1))
ax.set_xlim([0, 6])
ax.set_xticks(np.arange(7))
dist_plot = ax.set_xticklabels(['10', '100', '1k', '10k', '100k', '1M', '10M'])

目前各國確診/死亡案例 (柱狀圖)

將各國的資料根據確診數排序，排除確診人數小於 1000 的國家 (此時台灣確診 911、 死亡 8)，存在 top_country_df
下文確診案例、死亡案例、死亡率也一樣排除確診人數小於 1000 的國家

In [21]:
top_country_df = country_df.query('(date == @target_date) & (confirmed > 1000)').sort_values('confirmed', ascending=False)
        top_country_melt_df = pd.melt(top_country_df, id_vars='country', value_vars=['confirmed', 'fatalities'])
        fig = px.bar(top_country_melt_df.iloc[::-1],
                     x='value', y='country', color='variable', barmode='group',
                     title=f'Confirmed Cases/Deaths on {target_date}', text='value', height=1500, orientation='h')
        fig.show()

目前確診案例前 10 多的國家 (折線圖)

In [22]:
top10_countries = top_country_df.sort_values('confirmed', ascending=False).iloc[:10]['country'].unique()
top10_countries_df = country_df[country_df['country'].isin(top10_countries)]
fig = px.line(top10_countries_df,
              x='date', y='confirmed', color='country',
              title=f'Confirmed Cases for top 10 country as of {target_date}')
fig.show()

目前死亡案例前 10 多的國家 (折線圖)

In [23]:
top10_countries = top_country_df.sort_values('fatalities', ascending=False).iloc[:10]['country'].unique()
top10_countries_df = country_df[country_df['country'].isin(top10_countries)]
fig = px.line(top10_countries_df,
              x='date', y='fatalities', color='country',
              title=f'Fatalities for top 10 country as of {target_date}')
fig.show()

目前死亡率最高的前 30 個國家 (柱狀圖)

In [24]:
top_country_df = country_df.query('(date == @target_date) & (confirmed > 100)')
top_country_df['mortality_rate'] = top_country_df['fatalities'] / top_country_df['confirmed']
top_country_df = top_country_df.sort_values('mortality_rate', ascending=False)
fig = px.bar(top_country_df[:30].iloc[::-1],x='mortality_rate', y='country',title=f'Mortality rate HIGH: top 30 countries on {target_date}', text='mortality_rate', height=800, orientation='h')
fig.show()

目前死亡率最低的前 30 個國家 (柱狀圖)

In [25]:
fig = px.bar(top_country_df[:-30].iloc[::-1], x='mortality_rate', y='country',title=f'Mortality rate LOW: top 30 countries on {target_date}', text='mortality_rate', height=800, orientation='h')
fig.show()

宣告 all_country_df 紀錄每個國家最近一天的疫情資訊，並新增以下欄位:

  • 計算各國確診數的對數，記為 confirmed_log1p
  • 計算各國確診數的對數，記為 fatalities_log1p
  • 計算各國死亡率，記為 mortality_rate
In [26]:
all_country_df = country_df.query('date == @target_date')
all_country_df['confirmed_log1p'] = np.log10(all_country_df['confirmed'] + 1)
all_country_df['fatalities_log1p'] = np.log10(all_country_df['fatalities'] + 1)
all_country_df['mortality_rate'] = all_country_df['fatalities'] / all_country_df['confirmed']
all_country_df
Out[26]:
date country confirmed fatalities confirmed_log1p fatalities_log1p mortality_rate
72000 2021-01-31 Afghanistan 55023 2400 4.740552 3.380392 0.043618
72001 2021-01-31 Albania 78127 1380 4.892807 3.140194 0.017664
72002 2021-01-31 Algeria 107339 2891 5.030762 3.461198 0.026933
72003 2021-01-31 Andorra 9937 101 3.997299 2.008600 0.010164
72004 2021-01-31 Angola 19796 466 4.296599 2.669317 0.023540
... ... ... ... ... ... ... ...
72187 2021-01-31 Vietnam 1817 35 3.259594 1.556303 0.019263
72188 2021-01-31 West Bank and Gaza 158962 1833 5.201296 3.263399 0.011531
72189 2021-01-31 Yemen 2121 615 3.326745 2.789581 0.289958
72190 2021-01-31 Zambia 54217 763 4.734143 2.883093 0.014073
72191 2021-01-31 Zimbabwe 33388 1217 4.523603 3.085647 0.036450

192 rows × 7 columns

各國確診案例比較 (動態全球地圖) (以顏色區分等級)

顏色越深代表確診案例越多；反之則越少

In [27]:
fig = px.choropleth(all_country_df, locations="country", 
                    locationmode='country names', color="confirmed_log1p", 
                    hover_name="country", hover_data=["confirmed"],
                    range_color=[all_country_df['confirmed_log1p'].min(), all_country_df['confirmed_log1p'].max()], 
                    color_continuous_scale="peach", 
                    title='Countries with Confirmed Cases')
fig.show()

各國死亡案例比較 (動態全球地圖) (以顏色區分等級)

顏色越深代表死亡數越多；反之則越少

In [28]:
fig = px.choropleth(all_country_df, locations="country", 
                    locationmode='country names', color="fatalities_log1p", 
                    hover_name="country", range_color=[0, 4],
                    hover_data=['fatalities'],
                    color_continuous_scale="peach", 
                    title='Countries with fatalities')
fig.show()

各國死亡率比較 (動態全球地圖) (以顏色區分等級)

顏色越深代表死亡率越高；反之則越低

In [29]:
fig = px.choropleth(all_country_df, locations="country", 
                    locationmode='country names', color="mortality_rate", 
                    hover_name="country", range_color=[0, 0.12], 
                    color_continuous_scale="peach", 
                    title='Countries with mortality rate')
fig.show()

此時文中提到歐美澳的死亡率特別高，並提到一項假設是卡介苗在這些國家的接種率很高，也許跟卡介苗注射有關。不過考慮到這篇文章大概在 2020 年 5 到 6 月就有了，那時歐美的死亡率可能真的比較高，現在 (2021 年 2 月) 來看，歐美其實死亡率並沒有比較高。

死亡數演變 (折線圖) (取log) (超過 10 人死亡就開始記錄)

可以由這張圖比較各國的防疫能力 (醫療資源較充足、較早開始處理疫情的國家，曲線上升幅度會較緩慢)

In [30]:
n_countries = 10
n_start_death = 10
fatality_top_countires = top_country_df.sort_values('fatalities', ascending=False).iloc[:n_countries]['country'].values
country_df['date'] = pd.to_datetime(country_df['date'])
df_list = []
for country in fatality_top_countires:
    this_country_df = country_df.query('country == @country')
    start_date = this_country_df.query('fatalities > @n_start_death')['date'].min()
    this_country_df = this_country_df.query('date >= @start_date')
    this_country_df['date_since'] = this_country_df['date'] - start_date
    this_country_df['fatalities_log1p'] = np.log10(this_country_df['fatalities'] + 1)
    this_country_df['fatalities_log1p'] -= this_country_df['fatalities_log1p'].values[0]
    df_list.append(this_country_df)

tmpdf = pd.concat(df_list)
tmpdf['date_since_days'] = tmpdf['date_since'] / pd.Timedelta('1 days')
fig = px.line(tmpdf, x='date_since_days', y='fatalities_log1p', color='country', title=f'Fatalities by country since 10 deaths, as of {target_date}')
fig.add_trace(go.Scatter(x=[0, 28], y=[0, 4], name='Double by 7 days', line=dict(dash='dash', color=('rgb(200, 200, 200)'))))
fig.add_trace(go.Scatter(x=[0, 56], y=[0, 4], name='Double by 14 days', line=dict(dash='dash', color=('rgb(200, 200, 200)'))))
fig.add_trace(go.Scatter(x=[0, 84], y=[0, 4], name='Double by 21 days', line=dict(dash='dash', color=('rgb(200, 200, 200)'))))
fig.show()

確診人數前十國每日新增確診案例 (折線圖)

In [31]:
country_df['prev_confirmed'] = country_df.groupby('country')['confirmed'].shift(1)
country_df['new_case'] = country_df['confirmed'] - country_df['prev_confirmed']
country_df['new_case'].fillna(0, inplace=True)
top10_country_df = country_df[country_df['country'].isin(top10_countries)]

fig = px.line(top10_country_df,
              x='date', y='new_case', color='country',
              title=f'DAILY NEW Confirmed cases by country')
fig.show()

各國疫情發展情況: 確診案例 (點狀動態圖)

In [32]:
country_df['date'] = country_df['date'].apply(str)
country_df['confirmed_log1p'] = np.log1p(country_df['confirmed'])
country_df['fatalities_log1p'] = np.log1p(country_df['fatalities'])
fig = px.scatter_geo(country_df, locations="country", locationmode='country names', 
                      color="confirmed", size='confirmed', hover_name="country", 
                      hover_data=['confirmed', 'fatalities'],
                      range_color= [0, country_df['confirmed'].max()], 
                      projection="natural earth", animation_frame="date", 
                      title='COVID-19: Confirmed cases spread Over Time', color_continuous_scale="portland")
fig.show()

由於所需的數據大於 512 KB ，超過 chart-stdio 的上限，這裡僅放截圖:

各國省分詳細資訊

提供詳細省分詳細資訊的僅有 8 個國家:
Australia, Canada, China, Denmark, France, Netherlands, US, UK

In [33]:
for country in countries:
    province = train.query('country == @country')['province'].unique()
    if len(province) > 1:       
        print(f'Country {country} has {len(province)} provinces: {province}\n')
Country Australia has 8 provinces: ['Australian Capital Territory' 'New South Wales' 'Northern Territory' 'Queensland' 'South Australia' 'Tasmania' 'Victoria' 'Western Australia']

Country Canada has 14 provinces: ['Alberta' 'British Columbia' 'Manitoba' 'New Brunswick' 'Newfoundland and Labrador' 'Northwest Territories' 'Nova Scotia' 'Nunavut' 'Ontario' 'Prince Edward Island' 'Quebec' 'Repatriated Travellers' 'Saskatchewan' 'Yukon']

Country China has 33 provinces: ['Anhui' 'Beijing' 'Chongqing' 'Fujian' 'Gansu' 'Guangdong' 'Guangxi' 'Guizhou' 'Hainan' 'Hebei' 'Heilongjiang' 'Henan' 'Hong Kong' 'Hubei' 'Hunan' 'Inner Mongolia' 'Jiangsu' 'Jiangxi' 'Jilin' 'Liaoning' 'Macau' 'Ningxia' 'Qinghai' 'Shaanxi' 'Shandong' 'Shanghai' 'Shanxi' 'Sichuan' 'Tianjin' 'Tibet' 'Xinjiang' 'Yunnan' 'Zhejiang']

Country Denmark has 3 provinces: ['Faroe Islands' 'Greenland' nan]

Country France has 12 provinces: ['French Guiana' 'French Polynesia' 'Guadeloupe' 'Martinique' 'Mayotte' 'New Caledonia' 'Reunion' 'Saint Barthelemy' 'Saint Pierre and Miquelon' 'St Martin' 'Wallis and Futuna' nan]

Country Netherlands has 5 provinces: ['Aruba' 'Bonaire, Sint Eustatius and Saba' 'Curacao' 'Sint Maarten' nan]

Country US has 54 provinces: ['Alabama' 'Alaska' 'Arizona' 'Arkansas' 'California' 'Colorado' 'Connecticut' 'Delaware' 'District of Columbia' 'Florida' 'Georgia' 'Guam' 'Hawaii' 'Idaho' 'Illinois' 'Indiana' 'Iowa' 'Kansas' 'Kentucky' 'Louisiana' 'Maine' 'Maryland' 'Massachusetts' 'Michigan' 'Minnesota' 'Mississippi' 'Missouri' 'Montana' 'Nebraska' 'Nevada' 'New Hampshire' 'New Jersey' 'New Mexico' 'New York' 'North Carolina' 'North Dakota' 'Ohio' 'Oklahoma' 'Oregon' 'Pennsylvania' 'Puerto Rico' 'Rhode Island' 'South Carolina' 'South Dakota' 'Tennessee' 'Texas' 'Utah' 'Vermont' 'Virgin Islands' 'Virginia' 'Washington' 'West Virginia' 'Wisconsin' 'Wyoming']

Country United Kingdom has 11 provinces: ['Anguilla' 'Bermuda' 'British Virgin Islands' 'Cayman Islands' 'Channel Islands' 'Falkland Islands (Malvinas)' 'Gibraltar' 'Isle of Man' 'Montserrat' 'Turks and Caicos Islands' nan]

分析美國疫情

因為需要額外登入到 kaggle 下載資料，我就不搬運了，有興趣的可以看這裡

分析歐洲疫情

In [34]:
# Ref: https://www.kaggle.com/abhinand05/covid-19-digging-a-bit-deeper
europe_country_list =list([
    'Austria','Belgium','Bulgaria','Croatia','Cyprus','Czechia','Denmark','Estonia','Finland','France','Germany','Greece','Hungary','Ireland',
    'Italy', 'Latvia','Luxembourg','Lithuania','Malta','Norway','Netherlands','Poland','Portugal','Romania','Slovakia','Slovenia',
    'Spain', 'Sweden', 'United Kingdom', 'Iceland', 'Russia', 'Switzerland', 'Serbia', 'Ukraine', 'Belarus',
    'Albania', 'Bosnia and Herzegovina', 'Kosovo', 'Moldova', 'Montenegro', 'North Macedonia'])

country_df['date'] = pd.to_datetime(country_df['date'])
train_europe = country_df[country_df['country'].isin(europe_country_list)]
#train_europe['date_str'] = pd.to_datetime(train_europe['date'])
train_europe_latest = train_europe.query('date == @target_date')
fig = px.choropleth(train_europe_latest, locations="country", 
                    locationmode='country names', color="confirmed", 
                    hover_name="country", range_color=[1, train_europe_latest['confirmed'].max()], 
                    color_continuous_scale='portland', 
                    title=f'European Countries with Confirmed Cases as of {target_date}', scope='europe', height=800)
fig.show()

歐洲國家確診案例 (折線圖)

In [35]:
train_europe_march = train_europe.query('date >= "2020-03-01"')
fig = px.line(train_europe_march, x='date', y='confirmed', color='country', title=f'Confirmed cases by country in Europe, as of {target_date}')
fig.show()

歐洲國家死亡案例 (折線圖)

In [36]:
fig = px.line(train_europe_march, x='date', y='fatalities', color='country', title=f'Fatalities by country in Europe, as of {target_date}')
fig.show()

歐洲每日新增確診案例 (折線圖)

In [37]:
train_europe_march['prev_confirmed'] = train_europe_march.groupby('country')['confirmed'].shift(1)
train_europe_march['new_case'] = train_europe_march['confirmed'] - train_europe_march['prev_confirmed']
fig = px.line(train_europe_march, x='date', y='new_case', color='country', title=f'DAILY NEW Confirmed cases by country in Europe')
fig.show()

分析亞洲疫情

In [38]:
country_df['date'] = pd.to_datetime(country_df['date'])
country_latest = country_df.query('date == @target_date')
fig = px.choropleth(country_latest, locations="country", locationmode='country names', color="confirmed", hover_name="country", range_color=[1, 1000000], 
                    color_continuous_scale='portland', title=f'Asian Countries with Confirmed Cases as of {target_date}', scope='asia', height=800)
fig.show()

亞洲國家每日新增確診案例 (折線圖)

In [39]:
top_asian_country_df = country_df[country_df['country'].isin(['China', 'Indonesia', 'Iran', 'Japan', 'Korea, South', 'Malaysia', 'Philippines','India', 'Bangladesh', 'Pakistan', 'Saudi Arabia', 'Turkey'])]
            fig = px.line(top_asian_country_df, x='date', y='new_case', color='country', title=f'DAILY NEW Confirmed cases Asia')
            fig.update_layout(yaxis_range=[0, 100000])
            fig.show()

哪些國家已經脫離疫情高峰

以下地圖顯示，黃色國家的比例很高，代表確診人數快速上升，處於疫情高峰；藍色和紫色國家的比率較低，代表從高峰期開始下降。

我們可以發現歐美似乎有漸漸脫離高峰期，東南亞正在疫情高峰，德國及澳洲似乎已將疫情控制

In [40]:
max_confirmed = country_df.groupby('country')['new_case'].max().reset_index()
country_latest = pd.merge(country_latest, max_confirmed.rename({'new_case': 'max_new_case'}, axis=1))
country_latest['new_case_peak_to_now_ratio'] = country_latest['new_case'] / country_latest['max_new_case']
recovering_country = country_latest.query('new_case_peak_to_now_ratio < 0.5')
major_recovering_country = recovering_country.query('confirmed > 100')

fig = px.choropleth(country_latest, locations="country", locationmode='country names', color="new_case_peak_to_now_ratio", hover_name="country", range_color=[0, 1], hover_data=['confirmed', 'fatalities', 'new_case', 'max_new_case'], title='Countries with new_case_peak_to_now_ratio')
fig.show()

全球疫情何時趨緩

利用 sigmoid 函數建立一個模型協助我們預測各國的疫情何時趨緩 (確診人數不再暴增)

待續…

CATALOG
    FEATURED TAGS
    en c++ csharp leetcode c linux note others dotnet azure azure functions python data structure docker stm32f tour graph theory unboxing apache codility linear algebra nginx os FreeRTOS computer organization java opencv sql windows
    FRIENDS