def pushhandler(self, yyyyddd, weather, baseyear, func, skip_on_missing):
"""
Returns an interator of (yyyy, value, ...).
"""
for yearresult in self.subapp.push(yyyyddd, weather):
year = yearresult[0]
result = yearresult[1]
# Should we base everything off this year?
if year == self.year2:
self.denomterms.append(result)
self.denom = np.mean(self.denomterms)
# Print out all past results, re-based
for pastresult in self.pastresults:
diagnostic.record(self.region, pastresult[0], 'baseline', self.denom)
yield [pastresult[0], func(pastresult[1], self.denom)] + list(pastresult[1:])
if self.denom is None:
# Keep track of this until we have a base
self.pastresults.append(yearresult)
if year >= self.year1:
self.denomterms.append(result)
else:
diagnostic.record(self.region, year, 'baseline', self.denom)
# calculate this and tack it on
yield [year, func(result, self.denom)] + list(yearresult[1:])
def generate(region, year, temps, **kw):
temps2 = self.weather_change(temps)
result = np.nansum(curve(temps2)) / len(temps2)
if diagnostic.is_recording():
diagnostic.record(region, year, 'avgv', float(np.nansum(temps2)) / len(temps2))
if not np.isnan(result):
yield (year, result)
if isinstance(curve, AdaptableCurve):
curve.update(year, temps) # XXX: pass the original temps (labor decision, generalize?)
def generate(region, year, temps, **kw):
coeffs = self.getter(region, year, temps, curve)
if len(temps) == len(coeffs):
result = np.sum(self.weather_change(region, temps).dot(coeffs))
else:
raise RuntimeError("Unknown format for temps: " + str(temps.shape) + " <> len " + str(coeffs))
if diagnostic.is_recording():
for ii in range(temps.shape[0]):
diagnostic.record(region, year, 'var-' + str(ii), temps[ii])
if not np.isnan(result):
yield (year, result)
if isinstance(curve, AdaptableCurve):
curve.update(year, temps) # Passing in original (not weather-changed data)
def generate(region, year, temps, **kw):
temps = self.weather_change(temps)
assert temps.shape[1] == len(curve.curr_curve.ccs), "%d <> %d" % (temps.shape[1], len(curve.curr_curve.ccs))
#result = np.nansum(np.dot(temps, curve.curr_curve.ccs)) / len(temps)
result = np.dot(np.sum(temps, axis=0), curve.curr_curve.ccs) / len(temps)
if diagnostic.is_recording():
sumtemps = np.sum(temps, axis=0) / len(temps)
for ii in range(temps.shape[1]):
diagnostic.record(region, year, 'avgtk_' + str(ii+1), sumtemps[ii])
if not np.isnan(result):
yield (year, result)
if isinstance(curve, AdaptableCurve):
curve.update(year, temps)
