Analyzing the COVID-19 situation in Japan

import pandas as pd
import matplotlib as mpl
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
from matplotlib.dates import MonthLocator, DateFormatter
import folium
import geopandas

covid_data = pd.read_csv("coviddata.csv")
covid_data

Date	ALL	Hokkaido	Aomori	Iwate	Miyagi	Akita	Yamagata	Fukushima	Ibaraki	Tochigi	Gunma	Saitama	Chiba	Tokyo	Kanagawa	Niigata	Toyama	Ishikawa	Fukui	Yamanashi	Nagano	Gifu	Shizuoka	Aichi	Mie	Shiga	Kyoto	Osaka	Hyogo	Nara	Wakayama	Tottori	Shimane	Okayama	Hiroshima	Yamaguchi	Tokushima	Kagawa	Ehime	Kochi	Fukuoka	Saga	Nagasaki	Kumamoto	Oita	Miyazaki	Kagoshima	Okinawa
1/16/2020	1	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
1/17/2020	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
1/18/2020	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
1/19/2020	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
1/20/2020	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…
4/28/2023	11452	886	155	99	229	141	117	173	173	150	165	567	530	1613	788	281	159	164	75	88	245	199	181	524	174	86	277	771	374	129	51	79	32	94	304	67	32	106	94	28	377	42	91	102	84	61	68	227
4/29/2023	13082	899	170	121	272	152	153	173	135	171	184	776	671	1923	984	347	199	140	70	106	253	255	205	568	137	135	268	981	435	113	70	69	32	108	269	120	33	89	115	27	446	45	86	91	96	67	68	255
4/30/2023	6754	435	64	47	131	44	89	97	225	78	63	261	285	976	567	123	53	78	30	76	218	120	184	206	142	33	116	460	326	85	69	29	14	62	368	31	5	62	27	8	187	12	23	29	73	22	36	85
5/1/2023	5033	339	69	53	97	32	69	71	47	97	53	279	233	837	462	59	74	76	25	57	119	132	71	176	36	20	160	346	104	51	41	55	11	28	116	36	32	36	45	9	108	9	30	38	46	24	26	99
5/2/2023	16972	1260	204	205	258	287	175	291	68	307	235	946	795	2604	1179	512	242	225	99	157	279	386	196	846	75	157	379	1353	418	164	64	98	71	124	130	101	73	148	150	54	600	98	130	137	109	73	128	382

covid_data['Date'] = pd.to_datetime(covid_data['Date'])

covid_data.plot(kind='line', 
                x='Date', 
                y='ALL',
                grid=True,
                colormap="autumn",
                figsize=(19,13))

plt.xticks(rotation=45)
plt.title('"Trend in the number of newly confirmed cases (daily)"')
plt.xlabel('Day (monthly intervals)')
plt.ylabel('Total confirmed COVID-19 cases')
plt.gca().xaxis.set_major_locator(mpl.dates.MonthLocator()) 
plt.gca().xaxis.set_major_formatter(mpl.dates.DateFormatter("%Y-%m"))
plt.gcf().autofmt_xdate() # In order for the ticklabels not to overlap. HOWEVER, if there IS an overlap, remove this code
major_ticks=np.arange(0, 280000, 10000) # Values selected for Y axis
plt.yticks(major_ticks, major_ticks/10000) # If this outputs a decimal i.e. 500.0 etc., then remove "major_ticks, " 
    # For plt.yticks(actual, replacement), we provide the actual tick (data) values
    # And then the "replacement" values
plt.gca().yaxis.set_major_formatter(plt.matplotlib.ticker.StrMethodFormatter('{x:,.0f}')) # Add commas to Y axis
plt.margins(x=0,y=0)
plt.gca().get_legend().remove()
plt.show()

death_data = pd.read_csv("deathdata.csv")
death_data['Date'] = pd.to_datetime(death_data['Date'])
death_data = death_data[['Date', 'ALL']]
death_data

Date	ALL
5/9/2020	0
5/10/2020	8
5/11/2020	22
5/12/2020	25
5/13/2020	19
…	…
4/28/2023	34
4/29/2023	27
4/30/2023	14
5/1/2023	8
5/2/2023	19

sns.set_theme(style="whitegrid")
f, ax = plt.subplots(figsize=(12, 12))
plt.xticks(rotation=45)
plt.title('"Trends in the number of Deaths"')
plt.xlabel('Day (monthly intervals)')
plt.ylabel('Total COVID-19 related deaths')
sns.despine(f, left=True, bottom=True)
sns.scatterplot(x="Date", y="ALL",
                hue="ALL", size="ALL",
                palette="ch:s=.2,r=.7",
                sizes=(4, 36), linewidth=0,
                data=death_data,
                legend=False,
                ax=ax)
major_ticks=np.arange(0, 510, 10)
plt.yticks(major_ticks/1)
plt.gca().xaxis.set_major_locator(mpl.dates.MonthLocator()) 
plt.gca().xaxis.set_major_formatter(mpl.dates.DateFormatter("%Y-%m"))
plt.show()

import leafmap.foliumap as leafmap

j = leafmap.Map(center=[35.6895, 139.6917], zoom=5)
cumulative_data = pd.read_csv("confirmed_cases_cumulative.csv")
cumulative_data = cumulative_data[['Prefecture', 'Total Number', 'Incidence Rate', 'X', 'Y']]
cumulative_data

j.add_title("Cumulative number of COVID-19 cases per prefecture",
            font_size="20px", 
            align="center") 

j.add_geojson('dataoki.geojson')
j.add_points_from_xy(cumulative_data, x="Y", y="X",
                    color_column='Incidence Rate',
                     icon_names=['ambulance'],
                    marker_colors = ['blue', 'red'])
                    

Map of cumulative number of Japan COVID-19 cases per prefecture

Project Goal:

To analyze the COVID-19 situation in Japan and identify key trends and insights using publicly available data, focusing on nonlinear regression and geographical analysis. Data Sources:

1. COVID-19 data: confirmed cases (infection rate), deaths (death rate), cumulative data; was collected from the official Japanese government COVID-19 website: https://covid19.mhlw.go.jp/en/ ¹
2. Geographical Data was collected from OpenStreetMap: https://www.openstreetmap.org/

Language, Modules and Libraries:

Python, Pandas, Numpy, Matplotlib, Seaborn, Folium, ipyleaflet, GeoJSON, Geopandas, Leafmap

Methodology:

Until May 2023 and at the time of conducting this project, data for COVID-19 was collected and shared daily by the Japan Ministry of Health, Labour and Welfare. This data was downloaded and wrangled before being loaded into separate pandas data frames for analysis and plotting. Nonlinear regression models were plotted into line and scatter plot charts to visualize the daily infection rate and daily death rate respectively, with the timeline for these data sets covering three years. In terms of visual geographical analysis, cumulative numbers for confirmed COVID-19 cases over the aforementioned time period for each prefecture were plotted onto the map of Japan. These were plotted as marker clusters which included the latitude and longitude values for each prefecture. Vector data of Japan was also created as a GeoJSON.

Analysis and conclusion:

What we can see from the chart for the infection rate is that, between 2021 and 2023 - during the colder months, mostly January and February - it had spiked, before slowly tapering off into March and April. Another spike can be seen in August and September, consistently from 2020 to 2022.

We can also see that the infection rate from 2022 onwards was mostly always above 20,000 cases; whereas in 2021, outside of the abovementioned spike period, it never rose above 10,000 cases.

With regards to the death rate, as per the chart, a daily count exceeding 500 was only observed once. In addition, we can see similar patterns when compared to the daily infection rate i.e., spikes in January and February as well as August and September.

Another aspect to note is that we can see a spike in the death - and to a lesser extent, the infection rate - for May and June 2021, but this pattern is not necessarily observed again for May and June 2022 relative to other periods during the year.

When reflecting on the above data, we can see similarities in the timings of the spikes for both the infection and death rates, signifying a strong correlation. Because the spikes for both had been happening congruently during the same periods over the years, the data suggests that ultimately, the higher number of infected individuals, the higher the death count. This can be significant to note because what the data does not suggest, are scenarios whereby people are infected and then die after significant time had elapsed, meaning COVID-19 related deaths had mostly occurred within one month of confirmation of COVID-19. Indeed, when exploring further, data from other findings indicated that the median time it takes from the first symptoms of COVID-19 to death was roughly 18 days. ^{2 3}

As per the Japan map data, prefectures with the highest cumulative numbers were those with higher urban density e.g., Tokyo and prefectures surrounding Tokyo; as well the prefecture in the colder region of the country i.e., Hokkaido prefecture. We can see that the map data supports this, with such prefectures categorized as high-prevalence areas (the condition for a high-prevalence area is 1,000,000 or more cumulative numbers).

It would have been interesting to see whether or not the numbers would repeat itself for August and September 2023, however because the Japan Ministry of Health, Labour and Welfare ceased recording its data as of May 2023, the possibility of verifying this is unlikely. Nonetheless, the historical data is there, to help preempt and/or mitigate any risk concerning COVID-19 for 2023 and beyond.

Footnotes

1. Note that the Japan Ministry of Health, Labour and Welfare had stopped compiling and updating its COVID-19 data as of May 2023 ↩

2. https://www.drugs.com/medical-answers/covid-19-symptoms-progress-death-3536264/ ↩

3. https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-021-01314-w ↩