def main(args):
# Initialise
args, cfg = initialise(args)
# Load data
data, features, _ = load_data(args.input + 'data.h5',
train=True,
background=True)
# Fill Tau21 profile
profile = fill_profile(data, VAR_TAU21)
# Fit profile
fit = ROOT.TF1('fit', 'pol1', *FIT_RANGE)
profile.Fit('fit', 'RQ0')
intercept_val, coef_val = fit.GetParameter(0), fit.GetParameter(1)
intercept_err, coef_err = fit.GetParError(0), fit.GetParError(1)
# Create scikit-learn transform
ddt = LinearRegression()
ddt.coef_ = np.array([coef_val])
ddt.intercept_ = np.array([-coef_val * FIT_RANGE[0]])
ddt.offset_ = np.array([coef_val * FIT_RANGE[0] + intercept_val])
print "Fitted function:"
print " intercept: {:7.4f} ± {:7.4f}".format(intercept_val, intercept_err)
print " coef: {:7.4f} ± {:7.4f}".format(coef_val, coef_err)
# Save DDT transform
saveclf(ddt, 'models/ddt/ddt.pkl.gz')
return 0
def main(args):
# Initialise
args, cfg = initialise(args)
# Load data
data, _, _ = load_data(
'data/' + args.input) #Train=True removed since we use the data file
# -------------------------------------------------------------------------
####
#### # Initialise Keras backend
#### initialise_backend(args)
####
#### # Neural network-specific initialisation of the configuration dict
#### initialise_config(args, cfg)
####
#### # Keras import(s)
#### from keras.models import load_model
####
#### # NN
#### from run.adversarial.common import add_nn
#### with Profile("NN"):
#### classifier = load_model('models/adversarial/classifier/full/classifier.h5')
#### add_nn(data, classifier, 'NN')
#### pass
# -------------------------------------------------------------------------
# Compute background efficiency at sig. eff. = 50%
eff_sig = 0.10
fpr, tpr, thresholds = roc_curve(data['signal'],
data[VAR],
sample_weight=data['weight'])
idx = np.argmin(np.abs(tpr - eff_sig))
print "Background acceptance @ {:.2f}% sig. eff.: {:.2f}% ({} > {:.2f})".format(
eff_sig * 100., (fpr[idx]) * 100., VAR,
thresholds[idx]) #changed from 1-fpr[idx]
#print "Signal efficiency @ {:.2f}% bkg. acc.: {:.2f}% ({} > {:.2f})".format(eff_sig * 100., (fpr[idx]) * 100., VAR, thresholds[idx]) #changed from 1-fpr[idx]
print "Chosen target efficiency: {:.2f}%".format(EFF)
# Filling profile
data = data[data['signal'] == 0]
profile_meas, (x, y, z) = fill_profile(data)
# Format arrays
X = np.vstack((x.flatten(), y.flatten()))
X = X.T
Y = z.flatten()
# Fit KNN regressor
print "debugging more: x.shape = ", X.shape, ", y.ndim = ", Y.ndim
knn = KNeighborsRegressor(weights='distance')
knn.fit(X, Y)
# Save KNN classifier
saveclf(knn, 'models/knn/knn_{:s}_{}_{}.pkl.gz'.format(VAR, EFF, MODEL))
return 0
def main(args):
# Initialise
args, cfg = initialise(args)
# Load data
data, features, _ = load_data(args.input + 'data.h5',
train=True,
background=True)
#variable = VAR_TAU21
variable = VAR_N2
#variable = VAR_DECDEEP
#variable = VAR_DEEP
# Fill variable profile
profile = fill_profile(data, variable)
# Fit profile
if variable == VAR_N2:
fit_range = FIT_RANGE_N2
elif variable == VAR_TAU21:
fit_range = FIT_RANGE_TAU21
elif variable == VAR_DECDEEP:
fit_range = FIT_RANGE_DECDEEP
elif variable == VAR_DEEP:
fit_range = FIT_RANGE_DEEP
else:
print "variable invalid"
return 0
fit = ROOT.TF1('fit', 'pol1', *fit_range)
profile.Fit('fit', 'RQ0')
intercept_val, coef_val = fit.GetParameter(0), fit.GetParameter(1)
intercept_err, coef_err = fit.GetParError(0), fit.GetParError(1)
# Create scikit-learn transform
ddt = LinearRegression()
ddt.coef_ = np.array([coef_val])
ddt.intercept_ = np.array([-coef_val * fit_range[0]])
ddt.offset_ = np.array([coef_val * fit_range[0] + intercept_val])
print "Fitted function:"
print " intercept: {:7.4f} ± {:7.4f}".format(intercept_val, intercept_err)
print " coef: {:7.4f} ± {:7.4f}".format(coef_val, coef_err)
# Save DDT transform
saveclf(ddt, 'models/ddt/ddt_{}.pkl.gz'.format(variable))
print "got to the end of main()"
return 0
def train(data, variable, bg_eff, signal_above=False):
# Filling profile
data = data[data['signal'] == 0]
profile_meas, (x, y, z) = fill_profile(data,
variable,
bg_eff,
signal_above=signal_above)
# Format arrays
X = np.vstack((x.flatten(), y.flatten())).T
Y = z.flatten()
# Fit KNN regressor
knn = KNeighborsRegressor(weights='distance')
knn.fit(X, Y)
# Save KNN classifier
saveclf(knn, 'models/knn/knn_{:s}_{:.0f}.pkl.gz'.format(variable, bg_eff))
def main(args):
# Initialise
args, cfg = initialise(args)
# Load data
data, _, _ = load_data(
'data/' +
args.input) #, Train=True) removed since we use the data file
# -------------------------------------------------------------------------
####
#### # Initialise Keras backend
#### initialise_backend(args)
####
#### # Neural network-specific initialisation of the configuration dict
#### initialise_config(args, cfg)
####
#### # Keras import(s)
#### from keras.models import load_model
####
#### # NN
#### from run.adversarial.common import add_nn
#### with Profile("NN"):
#### classifier = load_model('models/adversarial/classifier/full/classifier.h5')
#### add_nn(data, classifier, 'NN')
#### pass
# -------------------------------------------------------------------------
# Compute background efficiency at sig. eff. = 50%
eff_sig = 0.10
fpr, tpr, thresholds = roc_curve(data['signal'],
data[VAR],
sample_weight=data['TotalEventWeight'])
idx = np.argmin(np.abs(tpr - eff_sig))
print "Background acceptance @ {:.2f}% sig. eff.: {:.2f}% ({} > {:.2f})".format(
eff_sig * 100., (fpr[idx]) * 100., VAR,
thresholds[idx]) #changed from 1-fpr[idx]
#print "Signal efficiency @ {:.2f}% bkg. acc.: {:.2f}% ({} > {:.2f})".format(eff_sig * 100., (fpr[idx]) * 100., VAR, thresholds[idx]) #changed from 1-fpr[idx]
print "Chosen target efficiency: {:.2f}%".format(EFF)
# Filling profile
data = data[data['signal'] == 0]
profile_meas, (x, y, err) = fill_profile_1D(data)
# Format arrays
X = x.reshape(-1, 1)
weights = 1 / err
print X
# Fit KNN regressor
if 'knn1D' == FIT:
knn = KNeighborsRegressor(5, weights='distance')
knn.fit(X, y) #.predict(X)
elif 'knn1D_v2' in FIT:
knn = KNeighborsRegressor(5, weights='uniform')
knn.fit(X, y) #.predict(X)
elif 'knn1D_v3' in FIT:
knn = KNeighborsRegressor(2, weights='uniform')
knn.fit(X, y) #.predict(X)
elif 'knn1D_v4' in FIT:
knn = KNeighborsRegressor(3, weights='distance')
knn.fit(X, y) #.predict(X)
elif 'poly2' in FIT:
knn = make_pipeline(PolynomialFeatures(degree=2), Ridge())
knn.fit(X, y) #.predict(X)
#knn1 = PolynomialFeatures(degree=2)
#knn1.fit(X, y)
#X_poly = knn1.fit_transform(X)
#knn = LinearRegression() #fit_intercept=False)
#knn.fit(X_poly, y, weights)
#score = round(reg.score(X_poly, y), 4)
#coef = reg.coef_
#intercept = reg.intercept_
#print score, coef, intercept
#knn.fit(X, y)#.predict(X)
#print "Fit parameters: ", knn.transform(X).shape #get_feature_names() #get_params() #knn.coef_
elif 'poly3' in FIT:
knn = make_pipeline(PolynomialFeatures(degree=3), Ridge())
knn.fit(X, y) #.predict(X)
# Create scikit-learn transform
elif 'lin' in FIT:
knn = LinearRegression()
knn.fit(X, y, weights)
elif 'erf' in FIT:
knn, pcov = curve_fit(func, x, y, p0=[73, 0.0004, 2000])
print "ERF: ", knn
else:
print "Weird FIT type chosen"
#coef_val = np.polyfit(x, y, deg=1, w=weights)
#knn.coef_ = np.array([coef_val[0]])
#knn.intercept_ = np.array([coef_val[1]]) #[-coef_val[0] * FIT_RANGE[0]])
#knn.offset_ = np.array([coef_val[0] * FIT_RANGE[0] + coef_val[1]])
print "Fitted function:"
print " coef: {}".format(knn.coef_)
print " intercept: {}".format(knn.intercept_)
# Save DDT transform
saveclf(knn,
'models/knn/{}_{:s}_{}_{}.pkl.gz'.format(FIT, VAR, EFF, MODEL))
# Save fit parameters to a ROOT file
#TCoef = ROOT.TVector3(coef[0], coef[1], coef[2])
#outFile = ROOT.TFile.Open("models/{}_jet_ungrtrk500_eff{}_stat{}_data.root".format(FIT, EFF, MIN_STAT),"RECREATE")
#outFile.cd()
#TCoef.SetName("coefficients")
#TCoef.Write()
#outFile.Close()
return 0