def test_naf_plot_cumulative_hazard_bandwith_1(self, block):
data1 = np.random.exponential(5, size=(2000, 1)) ** 2
naf = NelsonAalenFitter()
naf.fit(data1)
naf.plot_hazard(bandwidth=5.0, iloc=slice(0, 1700))
self.plt.title("test_naf_plot_cumulative_hazard_bandwith_1")
self.plt.show(block=block)
return
module_survival_data['event'] = 1
groups = module_survival_data['module_name']
T = module_survival_data['duration_weeks']
E = module_survival_data['event']
from lifelines import NelsonAalenFitter
naf = NelsonAalenFitter()
bandwidth = 3.
for i, each in enumerate(list(module_survival_data['module_name'].unique())):
ix = (groups == each)
naf.fit(T[ix], event_observed=E[ix], label=each)
if i == 0:
ax = naf.plot_hazard(bandwidth=bandwidth, ci_show=False)
else:
ax = naf.plot_hazard(ax=ax, bandwidth=bandwidth, ci_show=False)
ax.set_title("Hazard function of different modules | bandwidth=%.1f" %
bandwidth)
# Survival curves for tools
import pandas
from datetime import datetime, timedelta
from lifelines import KaplanMeierFitter
data_path = '/home/parthae/Documents/Projects/TISS_Git/projects/data_collation/data/data_latest'
cg_data = pandas.read_csv(
data_path +
'/state_level/state_tools_data/ct_metrics_tools_March31st2019.csv')
cg_data['state'] = 'cg'
# PURPOSE
ax = plt.subplot()
for purpose in df_cox.PURPOSE.unique():
is_pur = (df_cox.PURPOSE == purpose)
naf.fit(T[is_pur], event_observed=E[is_pur], label=purpose)
naf.plot(ax=ax)
# univariate analysis: hazard fxn
# NOTE: no real distinction
b = 3
# ORIG_CHN
ax = plt.subplot()
for chn in df_cox.ORIG_CHN.unique():
is_chn = (df_cox.ORIG_CHN == chn)
naf.fit(T[is_chn], event_observed=E[is_chn], label=chn)
naf.plot_hazard(ax=ax, bandwidth=b)
# PURPOSE
ax = plt.subplot()
for purpose in df_cox.PURPOSE.unique():
is_pur = (df_cox.PURPOSE == purpose)
naf.fit(T[is_pur], event_observed=E[is_pur], label=purpose)
naf.plot_hazard(ax=ax, bandwidth=b)
############################################################
# Cox proportional: statsmodels
############################################################
formula = ('AGE ~ {0} + {1}'.format(
'+'.join(['ORIG_CHN', 'PURPOSE'] + ['NUM_BO', 'CSCORE_MN']),
'+'.join(['np.log({0})'.format(c) for c in ['ORIG_AMT', 'ORIG_VAL']])))
ax = naf.plot_hazard(bandwidth=bandwidth,figsize=(10,8))
naf.fit(T[~small_ans], event_observed=E[~small_ans], label= category2)
naf.plot_hazard(ax=ax, bandwidth=bandwidth,figsize=(10,8))
plt.legend(title = column_name,prop={'size': 12})
plt.title('Curve Smoothening and Comparing({})'.format(column_name),size=12)
# utility function for a plot generation to explain the smoothness and bandwidth factor visually
def smooth_analysis():
n = 10
while(n<=100):
naf.plot_hazard(bandwidth=n,figsize=(15,13),label=n)
n+=10
plt.tight_layout()
plt.legend(title= 'Bandwidth',prop={'size': 12})
# plotting the NA Estimator
plot_estimator()
# getting a visualization of bandwidth and smoothness of the NA Estimator
smooth_analysis()
# smoothing the NA Estimator
naf.plot_hazard(bandwidth=10,figsize=(12,10),title="Smoother NA Estimator",label="NA Estimator")
# smoothing and comparing according to the column's entry
plot_function('Male','Female',3,'gender')
plot_function('Yes','No',3,'PhoneService')
plot_function('Yes','No',3,'Partner')
plot_function('Yes','No',3,'Dependents')
plot_function('Yes','No',3,'PaperlessBilling')
