covid-19-analysis/Analyser.py

# -*- coding: utf-8 -*-
"""
Project: Analyse worldwide COVID-19 Data and provide graphs etc.

@author Patrick Müller
"""
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import tkinter as tk

"""
Fields in csv:
dateRep
day
month
year
cases
deaths
countriesAndTerritories
geoId
countryterritoryCode
popData2018
"""
from datetime import datetime, timedelta


class Analyser:
	def __init__(self):
		# Pandas Settings
		pd.set_option('display.max_row', 50)
		pd.set_option('display.max_column', 10)
	
		self.df = pd.read_csv('statsfile.csv')
		self.df['dateRep'] = pd.to_datetime(self.df['dateRep'], format='%d/%m/%Y')

	def getAvailableCountries(self):
		sorted = self.df.sort_values('countriesAndTerritories')
		return sorted['countriesAndTerritories'].unique()

	def getAvailableDates(self):
		retList = []
		for date in self.df['dateRep'].unique():
			# To only get the substring in the format YYYY-MM-DD
			retList.append(str(date)[:10])
		retList.sort()
		return retList

	def getCasesGraph(self, country, start_date='2019-12-31', end_date=datetime.now().strftime('%Y-%m-%d')) -> str:
		"""
		Get a graph with the absolute number of cases by day for the entered country
		:param country: The country you wish to get the graph for
		:param start_date: The start date of the graph
		:param end_date: The end date of the graph
		:return: The path for the picture of the graph
		"""
		if country in self.getAvailableCountries():
			fig = plt.figure()
			ax = fig.add_subplot(111)
			plt.title(('Total cases in ' + country))

			countryData = self.df[self.df['countriesAndTerritories'].isin([country])]
			mask = (countryData['dateRep'] >= start_date) & (countryData['dateRep'] <= end_date)
			countryTimeData = countryData.loc[mask]
			countryTimeData = countryTimeData.sort_values('dateRep')
			countryTimeData['totalCases'] = countryTimeData['cases'].cumsum()

			countryTimeData.plot(ax=ax, x='dateRep', y='totalCases')

			plt.show(block=True)
			filePath = ('graphs/casesGraph_' + country + '_' + datetime.now().strftime('%Y-%m-%d'))
			fig.savefig(filePath)

			return filePath
		else:
			print('Unknown country')
			return '-1'

	def getCaseIncreaseGraph(self, country, start_date='2019-12-31', end_date=datetime.now().strftime('%Y-%m-%d')) -> str:
		"""
		Get a graph with the daily increase number of cases for the entered country
		:param country: The country you wish to get the graph for
		:param start_date: The start date of the graph
		:param end_date: The end date of the graph
		:return: The path for the picture of the graph
		"""
		if country in self.getAvailableCountries():
			fig = plt.figure()
			ax = fig.add_subplot(111)
			plt.title(('Daily new cases in ' + country))

			countryData = self.df[self.df['countriesAndTerritories'].isin([country])]
			mask = (countryData['dateRep'] >= start_date) & (countryData['dateRep'] <= end_date)
			countryTimeData = countryData.loc[mask]

			countryTimeData.plot(ax=ax, x='dateRep', y='cases')

			plt.show(block=True)
			filePath = ('graphs/casesIncreaseGraph_' + country + '_' + datetime.now().strftime('%Y-%m-%d'))
			fig.savefig(filePath)

			return filePath
		else:
			print('Unknown country')
			return '-1'

	def getTotalCases(self, country, date=datetime.now().strftime('%Y-%m-%d')) -> int:
		"""
		Get the current total cases for the entered country
		:param country: The country you want the case number for. Access available countries via getAvailableCountries()
		:param date: The date for which the case number is returned. Standard is the current date. Format YYYY-MM-DD
		:return: The case number
		"""
		countryData = self.df[self.df['countriesAndTerritories'].isin([country])]
		mask = (countryData['dateRep'] <= date)
		countryTimeData = countryData.loc[mask]
		return countryTimeData['cases'].sum()

	def getDeathGraph(self, country, start_date='2019-12-31', end_date=datetime.now().strftime('%Y-%m-%d')) -> str:
		"""
				Get a graph with the absolute number of cases by day for the entered country
				:param country: The country you wish to get the graph for
				:param start_date: The start date of the graph
				:param end_date: The end date of the graph
				:return: The path for the picture of the graph
				"""
		if country in self.getAvailableCountries():
			fig = plt.figure()
			ax = fig.add_subplot(111)
			plt.title(('Total deaths in ' + country))

			countryData = self.df[self.df['countriesAndTerritories'].isin([country])]
			mask = (countryData['dateRep'] >= start_date) & (countryData['dateRep'] <= end_date)
			countryTimeData = countryData.loc[mask]
			countryTimeData = countryTimeData.sort_values('dateRep')
			countryTimeData['totalDeaths'] = countryTimeData['deaths'].cumsum()

			countryTimeData.plot(ax=ax, x='dateRep', y='totalDeaths')

			plt.show(block=True)
			filePath = ('graphs/deathsGraph_' + country + '_' + datetime.now().strftime('%Y-%m-%d'))
			fig.savefig(filePath)

			return filePath
		else:
			print('Unknown country')
			return '-1'

	def getDeathIncreaseGraph(self, country, start_date='2019-12-31', end_date=datetime.now().strftime('%Y-%m-%d')) -> str:
		"""
		Get a graph with the daily increase number of cases for the entered country
		:param country: The country you wish to get the graph for
		:param start_date: The start date of the graph
		:param end_date: The end date of the graph
		:return: The path for the picture of the graph
		"""
		if country in self.getAvailableCountries():
			fig = plt.figure()
			ax = fig.add_subplot(111)
			plt.title(('Daily new deaths in ' + country))

			countryData = self.df[self.df['countriesAndTerritories'].isin([country])]
			mask = (countryData['dateRep'] >= start_date) & (countryData['dateRep'] <= end_date)
			countryTimeData = countryData.loc[mask]

			countryTimeData.plot(ax=ax, x='dateRep', y='deaths')

			plt.show(block=True)
			filePath = ('graphs/deathsIncreaseGraph_' + country + '_' + datetime.now().strftime('%Y-%m-%d'))
			fig.savefig(filePath)

			return filePath
		else:
			print('Unknown country')
			return '-1'

	def getDailyDeathRateGraph(self, country, start_date='2019-12-31', end_date=datetime.now().strftime('%Y-%m-%d')) -> str:
		"""
		Get a graph with the daily increase number of cases for the entered country
		:param country: The country you wish to get the graph for
		:param start_date: The start date of the graph
		:param end_date: The end date of the graph
		:return: The path for the picture of the graph
		"""
		if country in self.getAvailableCountries():
			fig = plt.figure()
			ax = fig.add_subplot(111)
			plt.title(('Daily death rate in ' + country) + ' in %')

			countryData = self.df[self.df['countriesAndTerritories'].isin([country])]
			mask = (countryData['dateRep'] >= start_date) & (countryData['dateRep'] <= end_date)
			countryTimeData = countryData.loc[mask]
			countryTimeData = countryTimeData.sort_values('dateRep')
			countryTimeData['totalCases'] = countryTimeData['cases'].cumsum()
			countryTimeData['totalDeaths'] = countryTimeData['deaths'].cumsum()
			countryTimeData['deathRate'] = countryTimeData['totalDeaths'] / countryTimeData['totalCases'] * 100

			countryTimeData.plot(ax=ax, x='dateRep', y='deathRate')

			plt.show(block=True)
			filePath = ('graphs/dailyDeathRateGraph_' + country + '_' + datetime.now().strftime('%Y-%m-%d'))
			fig.savefig(filePath)

			return filePath
		else:
			print('Unknown country')
			return '-1'