PlotEmulatorVsAcutal.py

import math
import os
import sys

import cPickle as pickle
import matplotlib.lines as mlines
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from DataReader.DataLoader import DataLoader
from Emulator.Emulator import *

from Preprocessor.PipeLine import *
from Utilities.Utilities import *

if len(sys.argv) != 2:
    print(
        "Use this script by entering: python %s Training_file" %
        (sys.argv[0]))
    sys.exit()


"""
Use trained emulator
"""
with open(sys.argv[1], "rb") as buff:
    data = pickle.load(buff)

emulator = data["emulator"]
training_data = data["data"]
scales = data["scales"]
nuggets = data["nuggets"]

num_run = training_data.sim_data.shape[0]
num_excluded = 1
emulator_results = []
actual_results = []

for validation_run in xrange(0, num_run, num_excluded):

    sim_data = training_data.sim_data
    sim_para = training_data.sim_para

    sim_data = np.delete(
        sim_data, slice(validation_run, validation_run + num_excluded), 0
    )
    sim_para = np.delete(
        sim_para, slice(validation_run, validation_run + num_excluded), 0
    )

    exp_result = training_data.sim_data[validation_run: validation_run + num_excluded]
    exp_para = training_data.sim_para[validation_run: validation_run + num_excluded]

    emulator.ResetData(sim_para, sim_data)
    # emulator.SetCovariance(squared_exponential)
    emulator.SetScales(scales)
    emulator.SetNuggets(nuggets)
    emulator.StartUp()

    for para in exp_para:
        mean, var = emulator.Emulate(para)
        emulator_results.append(mean.flatten().tolist())
    for data in exp_result:
        actual_results.append(data.tolist())

plt.plot(emulator_results, actual_results, "ro")

# also plot the line x=y for comparison
xmin = min(actual_results + emulator_results)
xmax = max(actual_results + emulator_results)
plt.plot([xmin, xmax], [xmin, xmax], "--")
plt.xlabel("Emulator result")
plt.ylabel("Actual results")
plt.show()