bres_regions = np.array([ 'b_res_20p70>=%0.3f'%resolution_array[i]+'& b_res_20p70<%0.3f'%resolution_array[i+1] for i in range(len(resolution_array)-1)])
branch_names = [c.strip() for c in branch_names]
features = [c.strip() for c in features]
features_cat = [c.strip() for c in features_cat]
region_names =bres_regions
for i_r,region in enumerate(bres_regions):
cuts = base_cuts+'&'+region
print cuts
data_frame = (rpd.read_root(utils.IO.featuresName[0],treeName, columns = branch_names)).query(cuts)
X_features = preprocessing.set_features(treeName,branch_names,features,cuts)
X_features_cat = (preprocessing.set_features(treeName,branch_names,features_cat,cuts))
X_test_features = preprocessing.get_test_sample(pd.DataFrame(X_features),0.)
nTot,dictVar = postprocessing.stackFeaturesReg(data_frame,branch_names,5)
true_pt = nTot[:,dictVar['Jet_mcPt']]
predictions_pt_caterina = nTot[:,dictVar['Jet_pt_reg']]
reco_pt = nTot[:,dictVar['Jet_pt']]
rel_diff_caterina = true_pt/predictions_pt_caterina
rel_diff_noreg = true_pt/reco_pt
# outTags = ['quantile_minmax_02-07','quantile_04']
outTags = ['quantile_04']
X_predictions_compare = []
for num in range(len(outTags)):
outTag = outTags[num]
#features = 'Jet_pt,Jet_eta,nPVs,Jet_mt,Jet_leadTrackPt,Jet_leptonPtRel,Jet_leptonPt,Jet_leptonDeltaR,Jet_neHEF,Jet_neEmEF,Jet_vtxPt,Jet_vtxMass,Jet_vtx3dL,Jet_vtxNtrk,Jet_vtx3deL'.split(",") #same as Caterina
#target = 'Jet_mcPt'.split(",")
####cuts='(Jet_mcPt > 20) & (Jet_eta<2.5 & Jet_eta>-2.5)'
#cuts='(Jet_pt > 20) & (Jet_eta<2.5 & Jet_eta>-2.5) '
#########################################
branch_names = [c.strip() for c in branch_names]
features = [c.strip() for c in features]
target = [c.strip() for c in target]
#print branch_names
import pandas as pd
import root_pandas as rpd
X_features = preprocessing.set_features("tree", branch_names, features, cuts)
X_target = preprocessing.set_target("tree", branch_names, target, cuts)
X_features, X_target = preprocessing.randomize_ft(X_features, X_target)
X_train_target = preprocessing.get_training_sample(pd.DataFrame(X_target), 0.9)
X_train_features = preprocessing.get_training_sample(pd.DataFrame(X_features),
0.9)
X_test_target = preprocessing.get_test_sample(pd.DataFrame(X_target), 0.9)
X_test_features = preprocessing.get_test_sample(pd.DataFrame(X_features), 0.9)
#now no need to split in test and training because i am not using testing here
#X_train_target = preprocessing.get_test_sample(pd.DataFrame(X_target),0.) #0.8
#X_train_features = preprocessing.get_test_sample(pd.DataFrame(X_features),0.)
from sklearn.externals import joblib
