def calculate_trends():
output = {'case7': {}, 'death': {}, 'case10': {}}
si = state_info()
print("Average case doubling time in days for past 7 days of data")
for state in si.get_states():
data = state_historic_data(state)
latest_data = data.get_latest_n(7)
fit = case_growth_rate(latest_data)
output['case7'][str(state)] = str(doubling_time(fit))
print(state + "," + str(doubling_time(fit)))
print("Average death doubling time in days for past 7 days of data")
for state in si.get_states():
data = state_historic_data(state)
latest_data = data.get_latest_n(7)
fit = death_growth_rate(latest_data)
output["death"][str(state)] = str(doubling_time(fit))
print(state + "," + str(doubling_time(fit)))
print("Average case doubling time in days since 10th confirmed case")
for state in si.get_states():
data = state_historic_data(state)
latest_data = data.get_after_n_cases(10)
fit = death_growth_rate(latest_data)
output["case10"][str(state)] = str(doubling_time(fit))
print(state + "," + str(doubling_time(fit)))
return output
def plot_states_growth(states,
data_name="positive",
logarithmic=True,
threshold=100,
filename=None):
index = 0
si = state_info()
max_days = 0
max_data = 0
for state in states:
data_handle = state_historic_data(state)
data_points = data_handle.get_after_n_cases(threshold)
dates = range(len(data_points))
data = list(map(lambda x: x[data_name], data_points))
if max_days < len(data_points):
max_days = len(data_points)
if max_data < max(data):
max_data = max(data)
plt.plot(dates,
data,
"C" + str(index) + ".-",
label=si.get_name(state))
index += 1
for double_time in [1, 2, 3, 5, 10]:
d = max_days
c = threshold * (2**(max_days / double_time))
if c > max_data:
c = max_data
d = (numpy.log(c / threshold) / numpy.log(2)) * double_time
plt.plot([0, d], [threshold, c], color='#bbbbbb', linestyle="--")
if logarithmic:
ax = plt.gca()
ax.set_yscale("log")
plt.legend()
plt.title(
"Growth Rate Since 100th Case\n(dotted lines represent doubling every 1, 2, 3, 5, and 10 days)"
)
plt.ylabel("Cases")
plt.xlabel("Days Since 100th Case")
fig = plt.gcf()
fig.set_size_inches(10, 7)
plt.grid(b=True)
if filename is None:
plt.show()
else:
fig.savefig("output/" + filename, dpi=100)
def plot_mortality_rate(states, threshold=2.5, filename=None, days=0):
index = 0
si = state_info()
for state in states:
data_handle = state_historic_data(state)
if days == 0:
data_points = data_handle.data
else:
data_points = data_handle.get_latest_n(days)
dates = list(
map(lambda x: datetime.strptime(str(x['date']), "%Y%m%d").date(),
data_points))
death = numpy.array(
list((map(
lambda x: 0
if ('death' not in x or x['death'] is None) else x['death'],
data_points))))
positive = numpy.array(
list((map(
lambda x: 1 if ('positive' not in x or x['positive'] is None)
else x['positive'], data_points))))
mortality_rate = death / positive * 100
plt.plot(dates,
mortality_rate,
"C" + str(index) + ".-",
label=si.get_name(state))
index += 1
plt.legend()
plt.title("Mortality Rate Trend by State")
plt.ylabel("% of positive cases resulting in death")
plt.axhline(y=threshold, color='r', linestyle='--')
fig = plt.gcf()
fig.set_size_inches(10, 7)
plt.grid(b=True, which="both")
if filename is None:
plt.show()
else:
fig.savefig("output/" + filename, dpi=100)
str(social_distancing_factor * 100) +
"% social distancing",
plot_color="C0")
mpl.title(
"KY Predicted infections over time based on death count, assuming %.1f%% mortality\nUsing SIR model, β=%0.2f, γ=0.15"
% (mortality_rate * 100, b))
mpl.grid(b=True)
if logarithmic:
mpl.gca().set_yscale("log")
mpl.show()
mpl.close()
if __name__ == "__main__":
state = "KY"
si = state_info()
pop = si.get_population(state)
data = state_historic_data(state).get_latest_n(45)
bootstrap_date = datetime.datetime.strptime(str(data[0]['date']),
'%Y%m%d').date()
positive = list(
map(
lambda x: 0
if 'positive' not in x or x['positive'] is None else x['positive'],
data))
death = list(
map(
lambda x: 0
if 'death' not in x or x['death'] is None else x['death'], data))
recovered = list(
map(
def plot_states_trend(states,
data_name='positive',
trendline=True,
logarithmic=True,
pop_adjusted=False,
days=0,
filename=None):
index = 0
si = state_info()
for state in states:
data_handle = state_historic_data(state)
if days > 0:
data_points = data_handle.get_latest_n(days)
else:
data_points = data_handle.get_after_n_cases(1)
dates = list(
map(lambda x: datetime.strptime(str(x['date']), "%Y%m%d").date(),
data_points))
cases = list(
map(
lambda x: x[data_name]
if data_name in x and x[data_name] is not None else 0,
data_points))
if pop_adjusted:
population = si.get_population(state)
cases = list(map(lambda x: x / population * 10000, cases))
(r, a) = calculate_trends.weighted_exponential_fit(
numpy.arange(len(cases)), cases)
if trendline:
fit_y = numpy.exp(a) * numpy.exp(r * numpy.arange(len(cases)))
plt.plot(dates, fit_y, "C" + str(index) + "--")
plt.plot(dates,
cases,
"C" + str(index) + "o",
label=si.get_name(state) + " (rate=%0.4f)" % r)
else:
plt.plot(dates,
cases,
"C" + str(index) + ".-",
label=si.get_name(state) + " (rate=%0.4f)" % r)
index += 1
if logarithmic:
ax = plt.gca()
ax.set_yscale("log")
plt.legend()
title = "Trend in " + data_name.capitalize() + " Count by State, "
if days == 0:
title += "All Days"
else:
title += "Last " + str(days) + " Days"
if pop_adjusted:
title += " (population adjusted)"
plt.title(title)
ylabel = data_name + " count"
if pop_adjusted:
ylabel += " per 10,000 people"
plt.ylabel(ylabel)
fig = plt.gcf()
fig.set_size_inches(10, 7)
plt.grid(b=True)
if filename is None:
plt.show()
else:
fig.savefig("output/" + filename, dpi=100)
def update_data():
si = state_info()
for state in si.get_states():
data = get_state_historic(state)
save_data(state + "_historic", data)