train_set_ptsel_sig, train_set_ptsel_bkg = splitdataframe_sigbkg(
train_set_ptsel, var_signal)
vardistplot(train_set_ptsel_sig, train_set_ptsel_bkg, mylistvariablesall,
"plots")
scatterplot(train_set_ptsel_sig, train_set_ptsel_bkg, mylistvariablesx,
mylistvariablesy, "plots")
correlationmatrix(train_set_ptsel_sig, "plots", "signal")
correlationmatrix(train_set_ptsel_bkg, "plots", "background")
if (doStandard == 1):
X_train = GetDataFrameStandardised(X_train)
if (doPCA == 1):
n_pca = 9
X_train, pca = GetPCADataFrameAndPC(X_train, n_pca)
plotvariancePCA(pca, "plots")
if (dotraining == 1):
trainedmodels = fit(names, classifiers, X_train, y_train)
savemodels(names, trainedmodels, "output", suffix)
if (doimportance == 1):
importanceplotall(mylistvariables, names, trainedmodels, suffix)
if (docrossvalidation == 1):
df_scores = cross_validation_mse(names, classifiers, X_train, y_train, 10,
ncores)
plot_cross_validation_mse(names, df_scores, suffix)
if (doRoCLearning == 1):
confusion(mylistvariables, names, classifiers, suffix, X_train, y_train, 5)
train_set_ptsel_sig, train_set_ptsel_bkg = splitdataframe_sigbkg(
train_set, myvariablesy)
vardistplot(train_set_ptsel_sig, train_set_ptsel_bkg, mylistvariablesall,
plotdir)
scatterplot(train_set_ptsel_sig, train_set_ptsel_bkg, mylistvariablesx,
mylistvariablesy, plotdir)
correlationmatrix(train_set_ptsel_sig, plotdir, "signal")
correlationmatrix(train_set_ptsel_bkg, plotdir, "background")
if (doStandard == 1):
X_train = GetDataFrameStandardised(X_train)
if (doPCA == 1):
n_pca = 5
X_train, pca = GetPCADataFrameAndPC(X_train, n_pca)
plotvariancePCA(pca, plotdir)
if (activateScikitModels == 1):
classifiersScikit, namesScikit = getclassifiers()
classifiers = classifiers + classifiersScikit
names = names + namesScikit
if (activateKerasModels == 1):
classifiersDNN, namesDNN = getclassifiersDNN(len(X_train.columns))
classifiers = classifiers + classifiersDNN
names = names + namesDNN
if (dotraining == 1):
trainedmodels = fit(names, classifiers, X_train, y_train)
savemodels(names, trainedmodels, output, suffix)