# Third party imports
import seaborn as sns
import matplotlib.pyplot as plt
# Local application imports
from scifin import timeseries as ts
from scifin import statistics as st
from scifin import montecarlo as mc
from scifin import classifier as cl
# Generate random walk processes
mc1 = mc.generate_series(n=100,
names_base="RW-",
series_model=ts.random_walk,
start_date="2000-01-01",
end_date="2020-01-01",
frequency='M',
start_value=1.,
sigma=0.01)
# Plot
ts.multi_plot(mc1)
# Build their covariance matrix
cov1 = st.covariance_from_ts(list_ts=mc1, min_periods=10)
corr1 = st.covariance_to_correlation(cov1)
# Plot this correlation matrix
sns.heatmap(corr1, cmap='viridis')
plt.title("Initial correlation matrix")
plt.show()
import numpy as np
import seaborn as sns
from sklearn import cluster
# Local application imports
from scifin import timeseries as ts
from scifin import montecarlo as mc
from scifin import classifier as cl
# Generate two kinds of processes
N = 50
mc_MA = mc.generate_series(n=N,
names_base="MA-",
series_model=ts.moving_average,
start_date="2000-01-01",
end_date="2020-01-01",
frequency='M',
cst=1.,
order=2,
coeffs=[0.5, 0.2],
sigma=0.4)
mc_Garch = mc.generate_series(n=N,
names_base="Garch-",
series_model=ts.garch,
start_date="2000-01-01",
end_date="2020-01-01",
frequency='M',
cst=0.2,
order_a=1,
coeffs_a=[0.2],
order_sig=1,