def plot_cascade(self, vertical=True):
if vertical:
fig_size = (12, 12)
ax_size = [0.45, 0.05, 0.5, 0.9]
else:
fig_size = (16, 8)
ax_size = [0.05, 0.45, 0.9, 0.5]
df = sc.dcp(self.data)
cutoff = 200e3
fig = pl.figure(figsize=fig_size)
df.sort(col='icer', reverse=False)
DA_data = hp.arr(df['opt_spend'])
inds = sc.findinds(DA_data > cutoff)
DA_data = DA_data[inds]
DA_data /= 1e6
DA_labels = df['shortname'][inds]
npts = len(DA_data)
colors = sc.gridcolors(npts, limits=(0.25, 0.75))
x = np.arange(len(DA_data))
pl.axes(ax_size)
for pt in range(npts):
loc = x[pt:]
this = DA_data[pt]
start = sum(DA_data[:pt])
prop = 0.9
color = colors[pt]
amount = sum(DA_data[:pt + 1])
amountstr = '%0.1f' % amount
if vertical:
pl.barh(loc, width=this, left=start, height=prop, color=color)
pl.text(amount,
x[pt],
amountstr,
verticalalignment='center',
color=colors[pt])
else:
pl.bar(loc, height=this, bottom=start, width=prop, color=color)
pl.text(x[pt],
amount + 1,
amountstr,
horizontalalignment='center',
color=colors[pt])
if vertical:
pl.xlabel('Spending for optimized investment cascade')
pl.gca().set_yticks(x)
ticklabels = pl.gca().set_yticklabels(DA_labels)
else:
pl.ylabel('Optimized investment cascade')
pl.gca().set_xticks(x)
ticklabels = pl.gca().set_xticklabels(DA_labels, rotation=90)
for t, tl in enumerate(ticklabels):
tl.set_color(colors[t])
pl.gca().set_facecolor('none')
pl.title('Investment cascade')
return fig
def plot_spending(self, which=None):
if which is None: which = 'current'
if which == 'current':
colkey = 'spend'
titlekey = 'Current'
elif which == 'optimized':
colkey = 'opt_spend'
titlekey = 'Optimized'
else:
errormsg = '"which" not recognized: %s' % which
raise Exception(errormsg)
df = sc.dcp(self.data)
fig = pl.figure(figsize=(10, 6))
print('WARNING, number of entries is hard-coded')
max_entries = 11
df.sort(col=colkey, reverse=True)
DA_data = hp.arr(df[colkey])
plot_data = list(DA_data[:max_entries - 1])
plot_data.append(sum(DA_data[max_entries:]))
plot_data = np.array(plot_data) / 1e6
total_averted = (DA_data.sum())
data_labels = ['%0.1fm' % datum for datum in plot_data]
DA_labels = df['shortname']
plot_labels = list(DA_labels[:max_entries - 1])
plot_labels.append('Other')
colors = self._getcolors(plot_labels)
pl.axes([0.15, 0.1, 0.45, 0.8])
pl.pie(plot_data,
labels=data_labels,
colors=colors,
startangle=90,
counterclock=False,
radius=0.5,
labeldistance=1.03)
pl.gca().axis('equal')
pl.title("%s spending (total: %s)" %
(titlekey, format(int(round(total_averted)), ',')))
pl.legend(plot_labels, bbox_to_anchor=(1, 0.8))
pl.gca().set_facecolor('none')
return fig
def makepackage(self,
burdenset=None,
intervset=None,
frpwt=None,
equitywt=None,
verbose=True,
die=False):
''' Make results '''
# Handle inputs
if burdenset is not None:
self.burdenset = burdenset # Warning, name is used both as key and actual set!
if intervset is not None: self.intervset = intervset
if frpwt is None: frpwt = 0.25
if equitywt is None: equitywt = 0.25
self.frpwt = frpwt
self.equitywt = equitywt
burdenset = self.projectref().burden(key=self.burdenset)
intervset = self.projectref().interv(key=self.intervset)
intervset.parse() # Ensure it's parsed
colnames = intervset.colnames
# Create new dataframe
origdata = sc.dcp(intervset.data)
critical_cols = [
'active', 'shortname', 'unitcost', 'spend', 'icer', 'frp', 'equity'
]
df = sc.dataframe()
for col in critical_cols: # Copy columns over
df[col] = sc.dcp(origdata[colnames[col]])
df['parsedbc'] = sc.dcp(origdata['parsedbc']) # Since not named
df.filter_out(key=0, col='active', verbose=True)
# Calculate people covered (spending/unitcost)
df['coverage'] = hp.arr(
df['spend']) / (self.eps + hp.arr(df['unitcost']))
# Pull out DALYS and prevalence
df.addcol('total_dalys',
value=0) # Value=0 by default, but just to be explicit
df.addcol('max_dalys', value=0)
df.addcol('total_prevalence', value=0)
df.addcol('dalys_averted', value=0)
notfound = []
lasterror = None
for r in range(df.nrows):
theseburdencovs = df['parsedbc', r]
for burdencov in theseburdencovs:
key = burdencov[0]
val = burdencov[1] # WARNING, add validation here
try:
thisburden = burdenset.data.findrow(
key=key,
col=burdenset.colnames['cause'],
asdict=True,
die=True)
df['total_dalys',
r] += thisburden[burdenset.colnames['dalys']]
df['max_dalys',
r] += thisburden[burdenset.colnames['dalys']] * val
df['total_prevalence',
r] += thisburden[burdenset.colnames['prevalence']]
except Exception as E:
lasterror = E # Stupid Python 3
print('HIIII %s' % str(E))
print(type(df['total_dalys', r]))
print(type(df['max_dalys', r]))
print(type(df['total_prevalence', r]))
print(type(thisburden[burdenset.colnames['dalys']]))
print(type(thisburden[burdenset.colnames['prevalence']]))
notfound.append(key)
# Validation
if len(notfound):
errormsg = 'The following burden(s) were not found: "%s"\nError:\n%s' % (
notfound, str(lasterror))
raise hp.HPException(errormsg)
invalid = []
for r in range(df.nrows):
df['dalys_averted',
r] = df['spend', r] / (self.eps + df['icer', r])
if df['dalys_averted', r] > df['max_dalys', r]:
errormsg = 'Data input error: DALYs averted for "%s" greater than total DALYs (%0.0f vs. %0.0f); please reduce total spending, increase ICER, increase DALYs, or increase max coverage' % (
df['shortname', r], df['dalys_averted', r], df['max_dalys',
r])
df['dalys_averted', r] = df[
'max_dalys',
r] # WARNING, reset to maximum rather than give error if die=False
invalid.append(errormsg)
if len(invalid):
errors = '\n\n'.join(invalid)
if die: raise Exception(errors)
else: print(errors)
# To populate with optimization results and fixed spending
self.budget = hp.arr(df['spend']).sum()
df.addcol('opt_spend')
df.addcol('opt_dalys_averted')
df.addcol('fixed')
# Store colors
nintervs = df.nrows
colors = sc.gridcolors(nintervs + 2,
asarray=True)[2:] # Skip black and white
colordict = sc.odict()
for c, name in enumerate(df['shortname']):
colordict[name] = colors[c]
self.colordict = colordict
self.data = df # Store it
if verbose:
print(
'Health package %s recalculated from burdenset=%s and intervset=%s'
% (self.name, self.burdenset, self.intervset))
return None
def optimize(self, budget=None, frpwt=None, equitywt=None, verbose=False):
# Handle inputs
if budget is None: budget = self.budget
if frpwt is None: frpwt = self.frpwt
if equitywt is None: equitywt = self.equitywt
self.budget = budget
self.frpwt = frpwt
self.equitywt = equitywt
df = sc.dcp(self.data)
if verbose:
print('Optimization inputs:')
print(' Budget: %s' % self.budget)
print(' FRP weight: %s' % self.frpwt)
print(' Equity weight: %s' % self.equitywt)
# Do the processing
frpdata = sc.dcp(hp.arr(df['frp']))
frpdata += 1.0 - frpdata.min()
frpdata /= frpdata.max()
equitydata = sc.dcp(hp.arr(df['equity']))
equitydata += 1 - equitydata.min()
equitydata /= equitydata.max()
equitydata = 1.0 - equitydata
df['icerwt'] = (1 - frpwt -
equitywt) + frpdata * frpwt + equitydata * equitywt
df['benefit'] = (1.0 / (hp.arr(df['icer']) + self.eps)) * hp.arr(
df['icerwt'])
# Handle fixed budgets
remaining = sc.dcp(self.budget)
for r in range(df.nrows):
if df['fixed', r]:
remaining -= df['spend', r]
df['opt_spend', r] = df['spend', r]
df['opt_dalys_averted', r] = df['dalys_averted', r]
# Do the "optimization"
df.sort(col='benefit', reverse=True)
max_dalys = hp.arr(df['max_dalys'])
icers = hp.arr(df['icer'])
if verbose: print('Optimizing...')
for r in range(df.nrows):
if not df['fixed', r]:
max_spend = max_dalys[r] * icers[r]
if verbose:
print(
' row %s | remaining %s | name %s | icer %s | icerwt %s | benefit %s | max_dalys %s | max_spend %s'
% (r, remaining, df['shortname', r], df['icer', r],
df['icerwt', r], df['benefit',
r], max_dalys[r], max_spend))
if remaining >= max_spend:
remaining -= max_spend
df['opt_spend', r] = max_spend
df['opt_dalys_averted', r] = max_dalys[r]
else:
df['opt_spend', r] = remaining
df['opt_dalys_averted',
r] = max_dalys[r] * remaining / max_spend
remaining = 0
df.sort(col='shortname')
self.data = df
if verbose:
print('Optimization output:')
print(self.data)
return None
def plot(self, which=None, n=None, axsize=None, figsize=None):
'''
Create a bar plot of the top causes of burden. By default, plots the top
10 causes of DALYs.
Version: 2018sep27
'''
# Set labels
titles = {
'dalys': 'Top causes of DALYs',
'deaths': 'Top causes of mortality',
'prevalence': 'Most prevalent conditions'
}
# Handle options
if which is None: which = list(titles.keys())
if n is None: n = 10
if axsize is None: axsize = (0.65, 0.15, 0.3, 0.8)
if figsize is None: figsize = (7, 4)
barw = 0.8
# Pull out data
df = sc.dcp(self.data)
nburdens = df.nrows
colors = sc.gridcolors(nburdens + 2, asarray=True)[2:]
colordict = sc.odict()
for c, cause in enumerate(df[self.colnames['cause']]):
colordict[cause] = colors[c]
# Convert to list
if not isinstance(which, list):
asarray = False
whichlist = sc.promotetolist(which)
else:
asarray = True
whichlist = which
# Loop over each option (may only be one)
figs = []
for which in whichlist:
colname = self.colnames[which]
try:
thistitle = titles[which]
thisxlabel = colname
except Exception as E:
errormsg = '"%s" not found, "which" must be one of %s (%s)' % (
which, ', '.join(list(titles.keys())), str(E))
raise Exception(errormsg)
# Process data
df.sort(col=colname, reverse=True)
topdata = df[:n]
try:
barvals = hp.arr(topdata[colname])
except Exception as E:
for r in range(topdata.nrows):
try:
float(topdata[colname, r])
except Exception as E2:
if topdata[colname, r] in ['', None]:
errormsg = 'For cause "%s", the "%s" value is missing or empty' % (
topdata[self.colnames['cause'], r], colname)
else:
errormsg = 'For cause "%s", could not convert "%s" value "%s" to number: %s' % (
topdata[self.colnames['cause'], r], colname,
topdata[colname, r], str(E2))
raise Exception(errormsg)
errormsg = 'An exception was encountered, but could not be reproduced: %s' % str(
E)
raise Exception(errormsg)
barlabels = topdata[self.colnames['cause']].tolist()
# Figure out the units
largestval = barvals[0]
if largestval > 1e6:
barvals /= 1e6
unitstr = ' (millions)'
elif largestval > 1e3:
barvals /= 1e3
unitstr = ' (thousands)'
else:
unitstr = ''
# Create plot
fig = pl.figure(facecolor='none', figsize=figsize)
ax = fig.add_axes(axsize)
ax.set_facecolor('none')
yaxis = pl.arange(n, 0, -1)
for i in range(n):
thiscause = topdata[self.colnames['cause'], i]
color = colordict[thiscause]
pl.barh(yaxis[i],
barvals[i],
height=barw,
facecolor=color,
edgecolor='none')
ax.set_yticks(pl.arange(10, 0, -1))
ax.set_yticklabels(barlabels)
sc.SIticks(ax=ax, axis='x')
ax.set_xlabel(thisxlabel + unitstr)
ax.set_title(thistitle)
sc.boxoff()
figs.append(fig)
if asarray: return figs
else: return figs[0]