def perform_finance_exp_test():
timestamps_list = fetch_hawkes_bund_data()
one_day_movement_up = timestamps_list[0][0]
finance_train_times = one_day_movement_up[:round(
len(one_day_movement_up) * 0.8)]
finance_test_times = one_day_movement_up[
round(len(one_day_movement_up) * 0.8):]
decay_candidates = np.logspace(0, 6, 20)
selected_decay = try_and_choose_decay([[finance_train_times]],
decay_candidates)
model, exp_simulation = fit_exp_hawkes_and_simulate(
[[finance_train_times]], selected_decay,
round(np.max(finance_test_times)) + 10)
print("predicted: " + str(exp_simulation.timestamps))
print("real: " + str(finance_train_times))
predicted_in_test_period = exp_simulation.timestamps[0][
exp_simulation.timestamps[0] > finance_test_times[0]]
real_zero_extends = np.full(
(len(predicted_in_test_period) - len(finance_test_times)), 0)
test_df = pd.DataFrame(
{
'real': np.append(np.around(finance_test_times), real_zero_extends,
0),
'predict': np.around(predicted_in_test_period)
},
columns=['real', 'predict'])
test_df.to_csv('../results/finance_obtained_times.csv', index=False)
# evaulation_df = pd.DataFrame(columns=['name', 'dataset', "score_on_train"])
evaluation_df = pd.read_csv('../results/baseline_scores.csv')
evaluation_df.loc[len(evaluation_df)] = [
'HawkesExpKern', 'finance',
model.score(),
model.score([[finance_test_times]])
]
evaluation_df.to_csv('../results/baseline_scores.csv', index=False)
import numpy as np
from tick.dataset import fetch_hawkes_bund_data
from tick.hawkes import HawkesConditionalLaw
from tick.plot import plot_hawkes_kernel_norms
from tick.plot import plot_hawkes_kernels
from tick.base import TimeFunction
import seaborn as sns
sns.set()
import matplotlib.pyplot as plt
timestamps_list = fetch_hawkes_bund_data()
abscisses = np.linspace(0., 10., 200)
def R_X_realisation(element):
times = np.array(element)
times_N_plus = times[2]
times_N_moins = times[3]
time_axe = np.sort(np.concatenate((times_N_moins, times_N_plus)))
I_R_X = []