# we split the data in 2 sets
data1, data2 = data1.splitInTwo()
#data 1 -> training data
#data 2 -> data to predict
# parameters of the calibration
lim = 150 # if ecal + hcal > lim, ecalib = math.nan
n_neighbors_ecal_eq_0 = 2000 # number of neighbors for ecal = 0
n_neighbors_ecal_neq_0 = 250 # number of neighbors for ecal ≠ 0
energystep_ecal_eq_0 = 1
energystep_ecal_neq_0 = 5
try:
# We import the calibration
filename = "calibrations/KNNGaussianFitDirect_162Kpart_ecal_train_ecal_neq_0_min_0.292026728392_hcal_train_ecal_eq_0_min_1.00043606758_hcal_train_ecal_neq_0_min_1.00002634525_lim_150_n_neighbors_ecal_eq_0_2000_n_neighbors_ecal_neq_0_250.calibration"
KNNGFD = importCalib(filename)
except FileNotFoundError:
# We create the calibration
KNNGFD = data1.KNNGaussianFitDirect(
n_neighbors_ecal_eq_0=n_neighbors_ecal_eq_0,
n_neighbors_ecal_neq_0=n_neighbors_ecal_neq_0,
lim=lim)
KNNGFD.saveCalib()
classname = KNNGFD.classname
#plot 3D Training points
fig = plt.figure(1, figsize=(5, 4))
usplt.plot3D_training(data1)
#plt.show()
savefig(fig, directory, classname + "_plot3D_training.png")
plt.close()
# parameters of the calibration
n_neighbors_ecal_eq_0 = 2000
n_neighbors_ecal_neq_0 = 250
weights = 'gaussian'
algorithm = 'auto'
sigma = 5
lim = 150
energystep = 1
kind = 'cubic'
cut = 2
try:
# We import the calibration
filename = "calibrations/KNNGaussianCleaning_162Kpart_cut_2_lim_150_n_neighbors_ecal_eq_0_2000_n_neighbors_ecal_neq_0_250_sigma_5.calibration"
KNNGC = importCalib(filename)
except FileNotFoundError:
# We create the calibration
KNNGC = data1.KNNGaussianCleaning(n_neighbors_ecal_eq_0,
n_neighbors_ecal_neq_0, weights,
algorithm, sigma, lim, energystep, kind,
cut)
KNNGC.saveCalib()
classname = KNNGC.classname
#plot 3D Training points
fig = plt.figure(1, figsize=(6, 4))
usplt.plot3D_training(data1)
plt.show()
savefig(fig, directory, classname + "_plot3D_training.png")
#importation of simulated particles filename = 'charged_hadrons_100k.energydata' data1 = importData(filename) filename = 'prod2_200_400k.energydata' data2 = importData(filename) # we merge the 2 sets of data data1 = data1.mergeWith(data2) # we split the data in 2 sets data1, data2 = data1.splitInTwo() #data 1 -> training data #data 2 -> data to predict # We import LR filename = "calibrations/LinearRegression_162Kpart_lim_150_lim_max_80_lim_min_20.calibration" LinearRegression = importCalib(filename) # We import KNNGFD filename = "calibrations/KNNGaussianFitDirect_162Kpart_ecal_train_ecal_neq_0_min_0.292026728392_hcal_train_ecal_eq_0_min_1.00043606758_hcal_train_ecal_neq_0_min_1.00002634525_lim_150_n_neighbors_ecal_eq_0_2000_n_neighbors_ecal_neq_0_250.calibration" KNNGFD = importCalib(filename) # We import KNNGF filename = "calibrations/KNNGaussianFit_162Kpart_hcal_train_ecal_eq_0_min_1.00043606758_lim_150_n_neighbors_ecal_eq_0_2000_n_neighbors_ecal_neq_0_250.calibration" KNNGF = importCalib(filename) # We import KNNGC filename = "calibrations/KNNGaussianCleaning_162Kpart_cut_2_lim_150_n_neighbors_ecal_eq_0_2000_n_neighbors_ecal_neq_0_250_sigma_5.calibration" KNNGC = importCalib(filename) # We import KNN filename = "calibrations/KNN_162Kpart_lim_150_n_neighbors_ecal_eq_0_2000_n_neighbors_ecal_neq_0_250_recalibrated_False_sigma_5.calibration"
# we split the data in 2 sets
data1, data2 = data1.splitInTwo()
#data 1 -> training data
#data 2 -> data to predict
# parameters of the calibration
lim_min = 20
lim_max = 80
lim = 150
# file to save the pictures
directory = "pictures/testLinearRegression/"
try:
# We import the calibration
filename = "calibrations/LinearRegression_162Kpart_lim_150_lim_max_80_lim_min_20.calibration"
LinearRegression = importCalib(filename)
except FileNotFoundError:
# We create the calibration
LinearRegression = data1.LinearRegression(lim_min=20, lim_max=80, lim=150)
# We save the calibration
LinearRegression.saveCalib()
classname = LinearRegression.classname
#plot 3D Training points
fig = plt.figure(1, figsize=(6, 4))
usplt.plot3D_training(data1)
plt.show()
savefig(fig, directory, classname + "_plot3D_training.png")
plt.close()
#plot 3D surface calibration
data1 = importData(filename)
filename = 'prod2_200_400k.energydata'
data2 = importData(filename)
# we merge the 2 sets of data
data1 = data1.mergeWith(data2)
# we split the data in 2 sets
data1, data2 = data1.splitInTwo()
#data 1 -> training data
#data 2 -> data to predict
# parameters of the calibration
nbLego = 60
try:
filename = "calibrations/CalibrationLego_162Kpart_delta_2.66856677123_ecal_max_262.853826966_hcal_max_262.853826966_lim_150_nbLego_100.calibration"
CalibrationLego = importCalib(filename)
except FileNotFoundError:
# We create the calibration
CalibrationLego = data1.CalibrationLego(nbLego=nbLego)
CalibrationLego.saveCalib()
classname = CalibrationLego.classname
#plot 3D Training points
fig = plt.figure(1, figsize=(6, 4))
usplt.plot3D_training(data1)
plt.show()
savefig(fig, directory, classname + "_plot3D_training.png")
plt.close()
#plot 3D surface calibration
fig = plt.figure(1, figsize=(6, 4))
Developed by Samuel Niang
For IPNL (Nuclear Physics Institute of Lyon)
Script to understand how does LinearRegression works.
"""
import matplotlib.pyplot as plt
import pfcalibration.usualplots as usplt
from pfcalibration.tools import savefig
from pfcalibration.tools import importData, importCalib
# file to save the pictures
directory = "pictures/testLinearRegression/"
try:
filename = "calibrations/LinearRegression_162Kpart_lim_150_lim_max_80_lim_min_20.calibration"
LR = importCalib(filename)
except FileNotFoundError:
#importation of simulated particles
filename = 'charged_hadrons_100k.energydata'
data1 = importData(filename)
filename = 'prod2_200_400k.energydata'
data2 = importData(filename)
# we merge the 2 sets of data
data1 = data1.mergeWith(data2)
# we split the data in 2 sets
data1, data2 = data1.splitInTwo()
#data 1 -> training data
#data 2 -> data to predict
# parameters of the calibration
lim_min = 20