dts.append(dt)
return ids, dts
def convert2date(x, fmt="%Y-%m-%d %H:%M:%S"):
from datetime import datetime as dt
return dt.strptime(x.split(".")[0], fmt).date()
import datetime
from scipy.interpolate import interp1d as spline
if __name__ == "__main__":
d = data("./consensusPredictionData")
predictions = d.importPredictions()
d = data("./analysisData/")
quantiles = d.importQuantiles()
toaQids = d.timingOfApprovalQids()
toaPreds = predictions.loc[toaQids]
toaQuants = quantiles.loc[toaQids]
toaQuants["value"] = toaQuants["value"].astype("datetime64[ns]")
gd = data("./consensusPredictionData/")
indivForecasts = gd.importIndividualForecasts()
resolutions = indivForecasts.reset_index()
if legend:
leg = ax.legend(loc="upper center",
fontsize=10,
ncol=3,
bbox_to_anchor=(1.1, 1.35),
frameon=False,
handletextpad=0.075,
labelspacing=0.01,
columnspacing=0.15)
else:
ax.get_legend().remove()
if __name__ == "__main__":
gd = data("./consensusPredictionData/")
individualScores = gd.importIndividualScores().reset_index()
consensusScores = gd.importConsensusScores().reset_index()
consensusScores["standin_id"] = consensusScores.expert + 900
# rank scores
def rankforecasters(x):
x["rank"] = x.logscore.rank() / len(x)
return x
individualAndConsScores = individualScores.append(consensusScores)
individualAndConsRanks = individualAndConsScores.groupby(
["qid"]).apply(rankforecasters)
individualRanks = individualAndConsRanks.loc[
individualAndConsRanks.standin_id < 900]
