def test_dataframe():
def dfprint(label, val):
sc.colorize('cyan', f'\n{label}')
print(val)
return None
print('Testing dataframe:')
a = sc.dataframe(cols=['x', 'y'], data=[[1238, 2], [384, 5], [666, 7]])
dfprint('Create dataframe', a)
dfprint('Print out a column', a['x'])
dfprint('Print out a row', a[0])
dfprint('Print out an element', a['x', 0])
a[0] = [123, 6]
dfprint('Set values for a whole row', a)
a['y'] = [8, 5, 0]
dfprint('Set values for a whole column', a)
a['z'] = [14, 14, 14]
dfprint('Add new column', a)
a.addcol('m', [14, 15, 16])
dfprint('Alternate way to add new column', a)
a.rmcol('z')
dfprint('Remove a column', a)
a.pop(1)
dfprint('Remove a row', a)
a.append([555, 2, -1])
dfprint('Append a new row', a)
a.insert(1, [660, 3, -1])
dfprint('Insert a new row', a)
a.sort()
dfprint('Sort by the first column', a)
a.sort('y')
dfprint('Sort by the second column', a)
a.addrow([770, 4, -1])
dfprint('Replace the previous row and sort', a)
dfprint('Return the row starting with value "555"', a.findrow(555))
a.rmrow()
dfprint('Remove last row', a)
a.rmrow(123)
dfprint('Remove the row starting with element "123"', a)
p = a.pandas()
dfprint('Convert to pandas', p)
q = p.add(p)
dfprint('Do a pandas operation', q)
a.pandas(q)
dfprint('Convert back', a)
a.filtercols(['m', 'x'])
dfprint('Filter to columns m and x', a)
b = sc.dcp(a)
dfprint('Dataframe copying:', a == b)
return a
def loaddata():
print('Loading data...')
dataurl = 'https://raw.githubusercontent.com/rstudio/shiny-examples/master/120-goog-index/data/trend_data.csv'
rawdata = sc.wget(dataurl).splitlines()
data = []
for r,rawline in enumerate(rawdata):
line = rawline.split(',')
if r==0: # Read header
cols = line
else: # Read data
tag = line[0]
yearnum = convertdate(line[1], '%Y-%m-%dT%I:%M:%fZ')
value = float(line[2]) if r>0 else line[2]
data.append([tag, yearnum, value])
df = sc.dataframe(cols=cols, data=data)
return df
def common_interventions(region=None,
income=None,
byplatform=False,
max_entries=10,
entry=None,
label=''):
sc.heading('Top interventions figure')
if not byplatform:
category_list = interv_data['Category 1'].tolist()
else:
category_list = interv_data['Platform'].tolist()
categories = sorted(set(category_list))
if byplatform:
order = [0, 2, 1, 3]
categories = [categories[o] for o in order]
keycols = ['Short name', 'Category 1', 'Category 2', 'Platform', 'ICER']
df = sc.dataframe(cols=keycols + ['Percent'], nrows=len(interv_data))
for key in keycols:
df[key] = interv_data[key]
df['Percent'] = 0.0
df.sort('Short name')
nspends, nintervs = R[0]['alloc'].shape
all_counts = pl.zeros((nspends, nintervs))
include_counts = sc.dcp(all_counts)
for co, country in enumerate(R.keys()):
proceed = True
if region and country_data['who_region', co] != region:
proceed = False # Could use continue
if income and country_data['income_group', co] != income:
proceed = False # Could use continue
if proceed:
alloc = R[country]['alloc']
counts = pl.array(alloc > 0, dtype=float)
if entry is None:
entries = range(nspends)
else:
entries = [entry]
for i in entries:
for j in range(nintervs):
all_counts[i, j] += 1
include_counts[i, j] += counts[i, j]
for j in range(nintervs):
include = include_counts[:, j].sum()
total = all_counts[:, j].sum()
df['Percent', j] = include / total
# Ensure sorted!!
data = sc.odict().make(keys=categories, vals=[])
counts = sc.odict().make(keys=categories, vals=0)
df.sort(col='Percent', reverse=True)
for j in range(nintervs):
if byplatform:
this_category = df['Platform', j]
else:
this_category = df['Category 1', j]
if counts[this_category] < max_entries:
data[this_category].append(df[j].tolist())
counts[this_category] += 1
if dosave:
df.export('results/rapid_top-interventions.xlsx')
fig = pl.figure(figsize=(9, 17))
ax = fig.add_axes([0.5, 0.1, 0.45, 0.85])
count = 50
ticklocs = []
ticklabels = []
if not byplatform:
darkest = [
pl.array([0.5, 0.1, 0.0]),
pl.array([0.0, 0.1, 0.5]),
pl.array([0.5, 0.0, 0.5]),
pl.array([0.1, 0.5, 0.0]),
]
else:
darkest = [
(0.5, 0.2, 0.0),
(0.0, 0.2, 0.5),
(0.0, 0.5, 0.0),
(0.0, 0.5, 0.3),
]
darkest = darkest[::-1]
position = df.cols.index('Percent')
for k, key, vals in sc.odict(data).enumitems():
count -= 2
count2 = 0
pl.text(-2,
count,
key,
fontweight='bold',
horizontalalignment='right',
fontsize=6)
maxval = len(vals)
for row in vals:
count -= 1
count2 += 1
thiscolor = darkest[k] + (count2 /
(maxval * 2)) * pl.array([1, 1, 1])
ticklocs.append(count)
ticklabels.append(row[0])
percentage = float(row[position]) * 100 # WARNING, fragile!!
pl.barh(count, percentage, facecolor=thiscolor, edgecolor='none')
ax.set_yticks(ticklocs)
ax.set_yticklabels(ticklabels, fontsize=4)
ax.set_title(label)
pl.xlim([0, 100])
pl.xlabel('Frequency of inclusion of intervention in EUHC package (%)')
if dosave:
connector = '' if not label else '-'
pl.savefig(f'results/rapid_top-interventions{connector}{label}.png',
dpi=200)
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
import sciris as sc
count = 0
def dfprint(label, val):
global count
count += 1
sc.colorize('blue', '\n%s. ' % count + label)
print(val)
return None
print('Testing dataframe:')
a = sc.dataframe(cols=['x', 'y'], data=[[1238, 2], [384, 5], [666, 7]])
dfprint('Create dataframe', a)
dfprint('Print out a column', a['x'])
dfprint('Print out a row', a[0])
dfprint('Print out an element', a['x', 0])
a[0] = [123, 6]
dfprint('Set values for a whole row', a)
a['y'] = [8, 5, 0]
dfprint('Set values for a whole column', a)
a['z'] = [14, 14, 14]
dfprint('Add new column', a)
a.addcol('m', [14, 15, 16])
dfprint('Alternate way to add new column', a)
a.rmcol('z')
dfprint('Remove a column', a)
a.pop(1)