def test_models(version=(30,-1), resample_flag=False, patients = -1, printReg = True): means = dict(educ_from_12=4, born_from_1960=10) if patients == -1: patients = DataUtilities.getDictReadofPatientsFilled() #patients = getDictReadofPatientsFilled() # read the survey survey, complete = read_complete(version, patients) print DataUtilities.get_version(version) #compare_survey_and_complete(survey, complete) #print 'all respondents', survey.len() #print 'complete', complete.len() # run the models if printReg: regs = run_regression_and_print(survey, version=version, means=means) else: regs = run_regression(survey, version) return regs
def errorTotal():
means = {}
version = (30,-1)
cumDict = {}
patientDict = DataUtilities.getDictReadofPatients()
for i in range(100):
patients = DataUtilities.getDictReadofPatientsFilled(patientDict = patientDict)
#resample
sPatients = sample_wr(patients, 91)
regs = TCGAlogReg.test_models(version = version, patients = sPatients, printReg = False)
for reg in regs.regs:
cumulative_odds = reg.report_odds(means, printCum = False)
for name, odds, p in cumulative_odds:
if name in cumDict.keys():
cumDict[name].append(odds)
else:
cumDict[name] = [odds]
print "ErrorTotal"
#create table
print r"\begin{table}[h]"
print r"\begin{tabular}{|l|l|l|l|}"
print r"\hline"
print r"\textbf{Type} & \textbf{Odds} & \textbf{Odds Lower} & \textbf{Odds Upper}\\ \hline"
for key, oddsList in sorted(cumDict.items(), key=lambda e: sum(e[1])):
if key not in mutationDict.getMutationsGreaterThan(1):
average = sum(oddsList) / float(len(oddsList))
lower = getLowerConfidence(oddsList)
upper= getUpperLowerConfidence(oddsList)
print key + " & " "{:.3f}".format(average) + " & " + "{:.3f}".format(lower) + " & " "{:.3f}".format(upper) + r"\\ \hline"
print "\end{tabular}"
print "\end{table}"
print ""
print r"\begin{table}[h]"
print r"\begin{tabular}{|l|l|l|l|}"
print r"\hline"
print r"\textbf{Type} & \textbf{Odds} & \textbf{Odds Lower} & \textbf{Odds Upper}\\ \hline"
names = []
averages = []
lowers = []
uppers = []
for key, oddsList in sorted(cumDict.items(), key=lambda e: sum(e[1])):
if key in mutationDict.getMutationsGreaterThan(1):
average = sum(oddsList) / float(len(oddsList))
lower = getLowerConfidence(oddsList)
upper= getUpperConfidence(oddsList)
print key + " & " "{:.3f}".format(average) + " & (" + "{:.3f}".format(lower) + ", " "{:.3f}".format(upper) + r")\\ \hline"
names.append(key)
averages.append(average)
lowers.append(lower)
uppers.append(upper)
print "\end{tabular}"
print "\caption{The above table shows the effect each mutation has when tested in conjunction with the other mutations}"
print "\end{table}"
plotRegMult(names, averages, lowers, uppers, "Mutation Log Odds With Total Error","Mutations", "Log Odds")
def errorMissing():
means = {}
version = (30,-1)
cumDict = {}
patientDict = DataUtilities.getDictReadofPatients()
for i in range(100):
patients = DataUtilities.getDictReadofPatientsFilled(patientDict = patientDict)
regs = TCGAlogReg.test_models(version = version, patients = patients, printReg = False)
for reg in regs.regs:
cumulative_odds = reg.report_odds(means, printCum = False)
for name, odds, p in cumulative_odds:
if name in cumDict.keys():
cumDict[name].append(odds)
else:
cumDict[name] = [odds]
print "ErrorMissing"
#create table
print r"\begin{table}[h]"
print r"\begin{tabular}{|l|l|l|}"
print r"\hline"
print r"\textbf{Type} & \textbf{Odds} & \textbf{Confidence Interval}\\ \hline"
namesL = []
averagesL = []
lowersL = []
uppersL = []
for key, oddsList in sorted(cumDict.items(), key=lambda e: sum(e[1])):
if key not in mutationDict.getMutationsGreaterThan(1):
average = sum(oddsList) / float(len(oddsList))
lower = getLowerConfidence(oddsList)
upper= getUpperConfidence(oddsList)
print key + " & " "{:.3f}".format(average) + " & (" + "{:.3f}".format(lower) + ", " "{:.3f}".format(upper) + r")\\ \hline"
namesL.append(key)
averagesL.append(average)
lowersL.append(lower)
uppersL.append(upper)
print "\end{tabular}"
print "\end{table}"
plotRegMult(namesL, averagesL, lowersL, uppersL, "Life Factor Log Odds With Sampling Error","Factor", "Log Odds")
print ""
print r"\begin{table}[h]"
print r"\begin{tabular}{|l|l|l|}"
print r"\hline"
print r"\textbf{Type} & \textbf{Odds} & \textbf{Confidence Interval}\\ \hline"
namesM = []
averagesM = []
lowersM = []
uppersM = []
for key, oddsList in sorted(cumDict.items(), key=lambda e: sum(e[1])):
if key in mutationDict.getMutationsGreaterThan(1):
average = sum(oddsList) / float(len(oddsList))
lower = getLowerConfidence(oddsList)
upper= getUpperConfidence(oddsList)
print key + " & " "{:.3f}".format(average) + " & (" + "{:.3f}".format(lower) + ", " "{:.3f}".format(upper) + r")\\ \hline"
namesM.append(key)
averagesM.append(average)
lowersM.append(lower)
uppersM.append(upper)
print "\end{tabular}"
print "\caption{The above table shows the effect each mutation has when tested independently the other mutations. The confidence intervals are shown for the error due to missing data.}"
print "\end{table}"
plotRegMult(namesM, averagesM, lowersM, uppersM, "Mutation Log Odds With Sampling Error","Mutation", "Log Odds")
plotRegMult(namesL + namesM, averagesL + averagesM, lowersL + lowersM, uppersL + uppersM, "Log Odds With Sampling Error","Factor", "Log Odds")
