def plot_fits(m):
# set font
book_graphics.set_font()
pl.figure(**book_graphics.three_quarter_plus_page_params)
for ii in range(2):
pl.subplot(1, 2, ii+1)
model = m[ii]
#
graphics.plot_data_bars(model.input_data, color='grey', label='Simulated data')
#
pl.plot(model.ages, model.pi_age_true, 'w-', linewidth=3)
pl.plot(model.ages, model.pi_age_true, 'k--', label='Truth')
#
#pl.plot(model.ages, model.vars['mu_age'].trace().T, '-', color='grey', alpha=.1)
pl.plot(model.ages, model.vars['mu_age'].stats()['mean'], 'w-', linewidth=4)
pl.plot(model.ages, model.vars['mu_age'].stats()['mean'], 'k-', linewidth=2, label='Posterior mean')
pl.plot(model.ages, model.vars['mu_age'].stats()['95% HPD interval'][:,0], 'k-', label='Uncertainty interval')
pl.plot(model.ages, model.vars['mu_age'].stats()['95% HPD interval'][:,1], 'k-')
#
if ii == 0:
pl.legend(fancybox=True, shadow=True, loc='upper center', bbox_to_anchor=(1.15,-.2), ncol=2)
pl.xlabel('Age (years)')
pl.ylabel('Rate (per PY)')
pl.axis([-5, 105, -.05, 1.])
pl.yticks([0, .25, .5, .75])
pl.xticks(fontsize='large')
book_graphics.subtitle('(%s)'%'ab'[ii])
#pl.subplots_adjust(top=.87, bottom=.44, wspace=.35) # l b r t w h #(top=.93, bottom=.53, wspace=.35)
pl.subplots_adjust(top=.9, bottom=.3, wspace=.35)
def plot_fits(m):
# set font
book_graphics.set_font()
pl.figure(**book_graphics.three_quarter_plus_page_params)
for ii in range(2):
pl.subplot(1, 2, ii + 1)
model = m[ii]
#
graphics.plot_data_bars(model.input_data,
color='grey',
label='Simulated data')
#
pl.plot(model.ages, model.pi_age_true, 'w-', linewidth=3)
pl.plot(model.ages, model.pi_age_true, 'k--', label='Truth')
#
#pl.plot(model.ages, model.vars['mu_age'].trace().T, '-', color='grey', alpha=.1)
pl.plot(model.ages,
model.vars['mu_age'].stats()['mean'],
'w-',
linewidth=4)
pl.plot(model.ages,
model.vars['mu_age'].stats()['mean'],
'k-',
linewidth=2,
label='Posterior mean')
pl.plot(model.ages,
model.vars['mu_age'].stats()['95% HPD interval'][:, 0],
'k-',
label='Uncertainty interval')
pl.plot(model.ages,
model.vars['mu_age'].stats()['95% HPD interval'][:, 1], 'k-')
#
if ii == 0:
pl.legend(fancybox=True,
shadow=True,
loc='upper center',
bbox_to_anchor=(1.15, -.2),
ncol=2)
pl.xlabel('Age (years)')
pl.ylabel('Rate (per PY)')
pl.axis([-5, 105, -.05, 1.])
pl.yticks([0, .25, .5, .75])
pl.xticks(fontsize='large')
book_graphics.subtitle('(%s)' % 'ab'[ii])
#pl.subplots_adjust(top=.87, bottom=.44, wspace=.35) # l b r t w h #(top=.93, bottom=.53, wspace=.35)
pl.subplots_adjust(top=.9, bottom=.3, wspace=.35)
return model
best_model = load_new_model()
output = pandas.read_csv('/home/j/Project/dismod/gbd/data/applications-data_af.csv')
# figure af-data
pl.figure(**book_graphics.half_page_params)
pl.subplot(1,2,1)
dismod3.graphics.plot_data_bars(best_model.get_data('p'))
pl.xlabel('Age (years)')
pl.ylabel('Prevalence (%)')
pl.yticks([0, .05, .1, .15, .2], [0, 5, 10, 15, 20])
my_axis(.22)
book_graphics.subtitle('(a)')
pl.subplot(1,2,2)
dismod3.graphics.plot_data_bars(best_model.get_data('i'))
pl.xlabel('Age (years)')
pl.ylabel('Incidence (per 1000 PY)')
pl.yticks([0, .001, .002, .003, .004], [0, 1, 2, 3, 4,])
my_axis(.0045)
book_graphics.subtitle('(b)')
pl.subplots_adjust(wspace=.35, top=.99, bottom=.14)
pl.savefig('book/graphics/af-data.pdf')
pl.savefig('book/graphics/af-data.png')
model = dismod3.data.load('/home/j/Project/dismod/dismod_status/prod/dm-37005')
#model.input_data = model.input_data.drop(['x_health_system_access'], axis=1)
return model
# figure cirrhosis-data
best_model = load_new_model()
pl.figure(**book_graphics.half_page_params)
pl.subplot(1,2,1)
dismod3.graphics.plot_data_bars(best_model.get_data('p'))
pl.xlabel('Age (years)')
pl.ylabel('Prevalence (%)')
pl.yticks([0, .001, .002, .003, .004], [0, 0.1, 0.2, 0.3, 0.4])
my_axis(.0045)
book_graphics.subtitle('(a)')
pl.subplot(1,2,2)
dismod3.graphics.plot_data_bars(best_model.get_data('pf'))
pl.xlabel('Age (years)')
pl.ylabel('Cause-specific mortality \n (per 10,000 PY)'+'\n\n', ha='center')
pl.yticks([0, .0008, .0016, .0024, .0032], [0, 8, 16, 24, 32])
my_axis(.0035)
pl.subplots_adjust(hspace=.35)
pl.subplots_adjust(wspace=.35)
book_graphics.subtitle('(b)')
pl.subplots_adjust(wspace=.35, top=.99, bottom=.14)
best_model = load_new_model()
output = pandas.read_csv(
'/home/j/Project/dismod/gbd/data/applications-data_af.csv')
# figure af-data
pl.figure(**book_graphics.half_page_params)
pl.subplot(1, 2, 1)
dismod3.graphics.plot_data_bars(best_model.get_data('p'))
pl.xlabel('Age (years)')
pl.ylabel('Prevalence (%)')
pl.yticks([0, .05, .1, .15, .2], [0, 5, 10, 15, 20])
my_axis(.22)
book_graphics.subtitle('(a)')
pl.subplot(1, 2, 2)
dismod3.graphics.plot_data_bars(best_model.get_data('i'))
pl.xlabel('Age (years)')
pl.ylabel('Incidence (per 1000 PY)')
pl.yticks([0, .001, .002, .003, .004], [
0,
1,
2,
3,
4,
])
my_axis(.0045)
book_graphics.subtitle('(b)')
mc.MAP(vars).fit(method='fmin_powell', tol=.00001, verbose=0)
#pl.plot(ages, vars['mu_age'].value, 'w', linewidth=3, **params)
if i == 0:
ax1 = fig.add_subplot(2,1,i+1)
ax1.plot(X, Y, 'ks', ms=4, mew=2)
ax1.plot(ages, vars['mu_age'].value, 'k', linewidth=2, linestyle=params['linestyle'])#, **params)
else:
ax2 = fig.add_subplot(2,1,i+1, sharex=ax1)
ax2.plot(X, Y, 'ks', ms=4, mew=2)
ax2.plot(ages, vars['mu_age'].value, 'k', linewidth=2, linestyle=params['linestyle'])#, **params)
pl.xlabel('$a$')
decorate_figure()
pl.setp(ax1.get_xticklabels(), visible=False)
book_graphics.subtitle(params['subt'] + ' ' + params['label'])
pl.subplots_adjust(hspace=.1, top=.99, bottom=.11)
pl.savefig('book/graphics/splines-fig.pdf')
# <codecell>
### @export 'spline_fig'
knots = range(0, 101, 5)
fig1_data = {}
for i, params in enumerate([dict(label=r'$\sigma = 0.5$', subt='(a)', smoothing=.5),
dict(label=r'$\sigma = 0.05$', subt='(b)', smoothing=.05),
dict(label=r'$\sigma = 0.005$', subt='(c)', smoothing=.005)]):
vars = age_pattern.age_pattern('t', ages=ages, knots=knots, smoothing=params.pop('smoothing'))
vars['mu_pred'] = mc.Lambda('mu_pred', lambda mu_age=vars['mu_age'], X=X : mu_age[X])
pl.plot(pl.arange(101),
pl.array(output['1k_' + params['type']]),
'k--',
linewidth=3,
label='{35}')
pl.plot(pl.arange(101),
pl.array(output['2k_' + params['type']]),
'k-',
linewidth=3,
label='{31, 35}')
pl.xlabel('Age (years)')
pl.ylabel(params['ylabel'] + '\n\n', ha='center')
pl.yticks(*params.get('yticks', ([0, .025, .05], [0, 2.5, 5])))
pl.axis(params.get('axis', [-5, 105, -.005, .06]))
book_graphics.subtitle(params['title'])
pl.subplots_adjust(top=.99, bottom=.14, wspace=.35, hspace=.25)
pl.legend(bbox_to_anchor=(.42, 0, .3, .53),
bbox_transform=pl.gcf().transFigure,
fancybox=True,
shadow=True,
title='Additional knots at:')
pl.savefig('book/graphics/oa_knee-knots.pdf')
pl.savefig('book/graphics/oa_knee-knots.png')
# figure oa_knee-i_prior
pl.figure(**book_graphics.three_quarter_page_params)
param_list = [
vars['mu_age'].value,
'k',
linewidth=2,
linestyle=params['linestyle']) #, **params)
else:
ax2 = fig.add_subplot(2, 1, i + 1, sharex=ax1)
ax2.plot(X, Y, 'ks', ms=4, mew=2)
ax2.plot(ages,
vars['mu_age'].value,
'k',
linewidth=2,
linestyle=params['linestyle']) #, **params)
pl.xlabel('$a$')
decorate_figure()
pl.setp(ax1.get_xticklabels(), visible=False)
book_graphics.subtitle(params['subt'] + ' ' + params['label'])
pl.subplots_adjust(hspace=.1, top=.99, bottom=.11)
pl.savefig('book/graphics/splines-fig.pdf')
# <codecell>
### @export 'spline_fig'
knots = range(0, 101, 5)
fig1_data = {}
for i, params in enumerate([
dict(label=r'$\sigma = 0.5$', subt='(a)', smoothing=.5),
dict(label=r'$\sigma = 0.05$', subt='(b)', smoothing=.05),
dict(label=r'$\sigma = 0.005$', subt='(c)', smoothing=.005)
]):
pl.figure(**book_graphics.half_page_params)
df = best_model.get_data('p')
pl.figure(**book_graphics.half_page_params)
for i in range(2):
pl.subplot(1,2,1+i)
dismod3.graphics.plot_data_bars(df[df['x_cv_past_year'] == i])
pl.xlabel('Age (years)')
pl.ylabel('Prevalence (%)')
pl.yticks([0, .07, .14, .21, .28], [0, 7, 14, 21, 28])
my_axis(.30)
if i == 0: book_graphics.subtitle('(a)')
if i == 1: book_graphics.subtitle('(b)')
pl.subplots_adjust(wspace=.35, top=.99, bottom=.14)
pl.savefig('book/graphics/anxiety-data_by_cv.pdf')
pl.savefig('book/graphics/anxiety-data_by_cv.png')
# figure anxiety-FE
data = pandas.read_csv('/home/j/Project/dismod/gbd/data/applications-data_anxiety.csv')
pl.figure(**book_graphics.full_page_params)
my_plot_data_bars(best_model.get_data('p'), color='grey', label='Period prevalence')
my_plot_data_bars(pm_model.get_data('p'), color='black',label='Point prevalence')
pl.plot(pl.arange(101), pl.array(data['best']), 'k-', linewidth=2, label='All data, with fixed effects')
pl.plot(pl.arange(101), pl.array(data['no_FE']), 'k--', linewidth=2, label='All data, without fixed effects')
def my_axis(ymax):
axis([-5, 105, -ymax / 10., ymax])
model = dismod3.data.load('/home/j/Project/dismod/notebooks/models/bipolar')
df = model.get_data('p')
pl.figure(**book_graphics.half_page_params)
for i in range(2):
pl.subplot(1, 2, 1 + i)
dismod3.graphics.plot_data_bars(df[df['x_cv_past_year'] == i])
pl.xlabel('Age (years)')
pl.ylabel('Prevalence (%)')
pl.yticks([0, .01, .02, .03, .04], [0, 1, 2, 3, 4])
pl.axis([-5, 105, -.0045, .045])
if i == 0: book_graphics.subtitle('(a) Past-year prevalence')
else: book_graphics.subtitle('(b) Past-month prevalence')
#title('cv_past_year = %d'%i)
pl.subplots_adjust(wspace=.35, bottom=.14)
pl.savefig('book/graphics/bipolar-data-by-cv.pdf')
# Compare alternative reference values
# -------------------------------------
model = dismod3.data.load('/home/j/Project/dismod/notebooks/models/bipolar')
#model.keep(areas=['super-region_0'], sexes=['male', 'total'], end_year=1997)
# remove expert prior on pyp effect
model.parameters['p']['fixed_effects'].pop('x_cv_past_year')
#dict(type='m_with', title='(d)', ylabel='With-condition mortality \n (per 100 PY)', yticks=([0, .1, .2, .3, .4], [0, 10, 20, 30, 40]), axis=[-5,105,-.045,.45], loc='upper left'),
]
for i, params in enumerate(param_list):
ax = pl.subplot(2,2,i+1)
dismod3.graphics.plot_data_bars(best_model.get_data(params['type']), color='grey')
pl.plot(pl.arange(101), pl.array(output['0k_'+params['type']]), 'k:', linewidth=3, label='{}')
pl.plot(pl.arange(101), pl.array(output['1k_'+params['type']]), 'k--', linewidth=3, label='{35}')
pl.plot(pl.arange(101), pl.array(output['2k_'+params['type']]), 'k-', linewidth=3, label='{31, 35}')
pl.xlabel('Age (years)')
pl.ylabel(params['ylabel']+'\n\n', ha='center')
pl.yticks(*params.get('yticks', ([0, .025, .05], [0, 2.5, 5])))
pl.axis(params.get('axis', [-5,105,-.005,.06]))
book_graphics.subtitle(params['title'])
pl.subplots_adjust(top=.99, bottom=.14, wspace=.35, hspace=.25)
pl.legend(bbox_to_anchor=(.42, 0, .3, .53), bbox_transform=pl.gcf().transFigure, fancybox=True, shadow=True, title='Additional knots at:')
pl.savefig('book/graphics/oa_knee-knots.pdf')
pl.savefig('book/graphics/oa_knee-knots.png')
# figure oa_knee-i_prior
pl.figure(**book_graphics.three_quarter_page_params)
param_list = [dict(type='p', title='(a)', ylabel='Prevalence (%)', yticks=([0, .12, .24, .36, .48], [0, 12, 24, 36, 48]), axis=[25,105,-.054,.54], loc='upper left'),
dict(type='i', title='(b)', ylabel='Incidence \n (per 1000 PY)', yticks=([0, .004, .008, .012, .016], [0, 4, 8, 12, 16]), axis=[-5,105,-.0018,.018], loc='upper left'),
#dict(type='f', title='(c)', ylabel='Excess mortality (Per 100 PY)', yticks=([0, .002, .004, .006, .008], [0, 2, 4, 6, 8]), axis=[-5,105,-.0018,.009], loc='upper left' ),
#dict(type='r', title='(c)', ylabel='Excess mortality (Per 100 PY)', yticks=([0, .002, .004, .006, .008], [0, 2, 4, 6, 8]), axis=[-5,105,-.0018,.009], loc='upper left' ),
def my_axis(ymax):
axis([-5,105,-ymax/10.,ymax])
model = dismod3.data.load('/home/j/Project/dismod/notebooks/models/bipolar')
df = model.get_data('p')
pl.figure(**book_graphics.half_page_params)
for i in range(2):
pl.subplot(1,2,1+i)
dismod3.graphics.plot_data_bars(df[df['x_cv_past_year'] == i])
pl.xlabel('Age (years)')
pl.ylabel('Prevalence (%)')
pl.yticks([0, .01, .02, .03, .04], [0, 1, 2, 3, 4])
pl.axis([-5,105,-.0045, .045])
if i == 0: book_graphics.subtitle('(a) Past-year prevalence')
else: book_graphics.subtitle('(b) Past-month prevalence')
#title('cv_past_year = %d'%i)
pl.subplots_adjust(wspace=.35, bottom=.14)
pl.savefig('book/graphics/bipolar-data-by-cv.pdf')
# Compare alternative reference values
# -------------------------------------
model = dismod3.data.load('/home/j/Project/dismod/notebooks/models/bipolar')
#model.keep(areas=['super-region_0'], sexes=['male', 'total'], end_year=1997)
# remove expert prior on pyp effect
model.parameters['p']['fixed_effects'].pop('x_cv_past_year')
model.parameters["f"]["parameter_age_mesh"] = [0, 100]
return model
csmr_model = load_new_model()
# figure alcohol-data
pl.figure(**book_graphics.full_plus_page_params)
pl.subplot(2, 2, 1)
dismod3.graphics.plot_data_bars(csmr_model.get_data("p"))
pl.xlabel("Age (years)")
pl.ylabel("Prevalence (%)")
pl.yticks([0, 0.03, 0.06, 0.09, 0.12], [0, 3, 6, 9, 12])
my_axis(0.13)
book_graphics.subtitle("(a)")
pl.subplot(2, 2, 2)
dismod3.graphics.plot_data_bars(csmr_model.get_data("i"))
pl.xlabel("Age (years)")
pl.ylabel("Incidence (per PY)")
pl.yticks([0, 3, 6, 9, 12])
my_axis(13)
book_graphics.subtitle("(b)")
pl.subplot(2, 2, 3)
dismod3.graphics.plot_data_bars(csmr_model.get_data("csmr"))
pl.xlabel("Age (years)")
pl.ylabel("Cause-specific mortality \n (per 100,000 PY)" + "\n\n", ha="center")
