def plot_data_trials(Nclasses,
X_train,
y_train,
n_trials_to_show=2,
colors=["r", "k"]):
# Nclasses = len(X_data_labels)
# print Nclasses
## Plotting evolution in time of several trials for both classes
gl.scatter_3D(0, 0, 0, nf=1, na=0)
for i in range(Nclasses):
X_train_class_i = [
X_train[j] for j in np.where(np.array(y_train) == i)[0]
]
for ntr in range(n_trials_to_show):
X_train_class_i_n = [X_train_class_i[ntr]]
X_train_class_i_n = np.concatenate(X_train_class_i_n, axis=0)
gl.scatter_3D(X_train_class_i_n[:, 0],
X_train_class_i_n[:, 1],
X_train_class_i_n[:, 2],
color=colors[i],
nf=0,
na=0,
labels=[
"Time Evolution of trials different classes",
"D1", "D2", "D3"
])
def plot_trials_for_same_instance(X_data_trials,
X_data_labels,
X_train,
y_train,
colors=["r", "k"],
time_show=100,
normalize=True):
# For a fixed time instant, the plotting of points from several trials of both classes
gl.scatter_3D(0, 0, 0, nf=1, na=0)
for i in range(len(X_data_trials)):
class_i = X_data_labels[i]
if (normalize == True):
caca = gf.normalize_module(X_data_trials[i][[time_show], :])
else:
caca = X_data_trials[i][[time_show], :]
caca = caca.flatten()
gl.scatter_3D(
caca[0],
caca[1],
caca[2],
nf=0,
na=0,
color=colors[class_i],
labels=[
"Trials for the same time instant for different classes", "D1",
"D2", "D3"
])
def plot_means(Nclasses, X_train, y_train,
colors = ["r","k"], normalize = True):
print Nclasses
## Get the time average profile of every label.
# For every label, we average across time to get the time profile.
# We kind of should assume that the trials are somewhat time-aligned
# X_data_ave = dp.get_timeSeries_average_by_label(X_All_labels, channel_sel = channel_sel)
X_data_ave = dp.get_average_from_train(Nclasses, X_train, y_train, normalize = normalize)
# Evolution of the means of each class in time in time representation
gl.plot([0],[0])
for i in range(1):
max_val = 0
if (i >= 1):
max_val += np.max(np.abs(X_data_ave[i-1])) + np.max(np.abs(X_data_ave[i]))
gl.plot([], X_data_ave[i] + max_val, color = colors[i], nf = 0,
labels = ["Mean value of the 70 Channels","Time Index","Channels"])
# Evolution of the means of each class in time in Spherical representation
gl.scatter_3D(0, 0,0, nf = 1, na = 0)
for i in range(Nclasses):
gl.scatter_3D(X_data_ave[i][:,0], X_data_ave[i][:,1],X_data_ave[i][:,2], color = colors[i],
nf = 0, na = 0, labels = ["Mean Time Evolution the different classes", "D1","D2","D3"])
def plot_trials_for_same_instance(X_data_trials, X_data_labels, X_train, y_train,
colors = ["r","k"], time_show = 100, normalize = True):
# For a fixed time instant, the plotting of points from several trials of both classes
gl.scatter_3D(0, 0,0, nf = 1, na = 0)
for i in range(len(X_data_trials)):
class_i = X_data_labels[i]
if (normalize == True):
caca = gf.normalize_module(X_data_trials[i][[time_show],:])
else:
caca = X_data_trials[i][[time_show],:]
caca = caca.flatten()
gl.scatter_3D(caca[0],caca[1],caca[2],
nf = 0, na = 0, color = colors[class_i],labels = ["Trials for the same time instant for different classes", "D1","D2","D3"])
def plot_data_trials(Nclasses, X_train, y_train,
n_trials_to_show = 2 , colors = ["r","k"]):
# Nclasses = len(X_data_labels)
# print Nclasses
## Plotting evolution in time of several trials for both classes
gl.scatter_3D(0, 0,0, nf = 1, na = 0)
for i in range(Nclasses):
X_train_class_i = [ X_train[j] for j in np.where(np.array(y_train) == i)[0]]
for ntr in range (n_trials_to_show):
X_train_class_i_n = [X_train_class_i[ntr]]
X_train_class_i_n = np.concatenate(X_train_class_i_n, axis = 0)
gl.scatter_3D(X_train_class_i_n[:,0], X_train_class_i_n[:,1],X_train_class_i_n[:,2], color = colors[i],
nf = 0, na = 0, labels = ["Time Evolution of trials different classes", "D1","D2","D3"])
def scatter_deltaDailyMagic(self):
## PLOTS DAILY HEIKE ASHI
ddelta = self.get_timeSeriesbyName("RangeCO")
hldelta = self.get_timeSeriesbyName("RangeHL")
mdelta = self.get_magicDelta()
labels = ["Delta Magic Scatter", "Magic", "Delta"]
gl.scatter(mdelta, ddelta, labels=labels, legend=[self.symbolID], nf=1)
# gl.set_subplots(1,1)
gl.scatter_3D(mdelta,
ddelta,
hldelta,
labels=labels,
legend=[self.symbolID],
nf=1)
def scatter_deltaDailyMagic(self):
## PLOTS DAILY HEIKE ASHI
ddelta = self.get_timeSeriesbyName("RangeCO")
hldelta = self.get_timeSeriesbyName("RangeHL")
mdelta = self.get_magicDelta()
labels = ["Delta Magic Scatter","Magic","Delta"]
gl.scatter(mdelta,ddelta,
labels = labels,
legend = [self.symbolID],
nf = 1)
# gl.set_subplots(1,1)
gl.scatter_3D(mdelta,ddelta, hldelta,
labels = labels,
legend = [self.symbolID],
nf = 1)
def plot_means(Nclasses, X_train, y_train, colors=["r", "k"], normalize=True):
print Nclasses
## Get the time average profile of every label.
# For every label, we average across time to get the time profile.
# We kind of should assume that the trials are somewhat time-aligned
# X_data_ave = dp.get_timeSeries_average_by_label(X_All_labels, channel_sel = channel_sel)
X_data_ave = dp.get_average_from_train(Nclasses,
X_train,
y_train,
normalize=normalize)
# Evolution of the means of each class in time in time representation
gl.plot([0], [0])
for i in range(1):
max_val = 0
if (i >= 1):
max_val += np.max(np.abs(X_data_ave[i - 1])) + np.max(
np.abs(X_data_ave[i]))
gl.plot(
[],
X_data_ave[i] + max_val,
color=colors[i],
nf=0,
labels=["Mean value of the 70 Channels", "Time Index", "Channels"])
# Evolution of the means of each class in time in Spherical representation
gl.scatter_3D(0, 0, 0, nf=1, na=0)
for i in range(Nclasses):
gl.scatter_3D(X_data_ave[i][:, 0],
X_data_ave[i][:, 1],
X_data_ave[i][:, 2],
color=colors[i],
nf=0,
na=0,
labels=[
"Mean Time Evolution the different classes", "D1",
"D2", "D3"
])
def get_2EM_vectors(X_All_labels,
label_classes,
max_trials=50,
channel_sel=None,
plot_flag=0):
Nclasses = len(X_All_labels)
# Instead of obtaining the average profile at each sime instant, we run
# a 2-EM clustering to each label, at each time instance.
# For each time instance,instead of getting the previous mean, we get 2 vectors.
# The mean vector of the class with high kappa and the one with low kappa
# We work on the assumtion that we have 2 clusters,
X_data_ave_EM_plus, X_data_ave_EM_minus = dp.get_labels_ave_EM(
X_All_labels,
label_classes,
max_trials=max_trials,
channel_sel=channel_sel)
# Get the 2-EM for all the trials !!
# For every trial we just run a 2 EM and get the direction of the
X_trials_EM_plus, X_trials_EM_minus = dp.get_X_trials_EM(
X_All_labels,
label_classes,
max_trials=max_trials,
channel_sel=channel_sel)
plot_flag = 0
if (plot_flag):
## For the 2-EM across time
gl.scatter_3D(0, 0, 0, nf=1, na=0)
for i in range(Nclasses):
gl.scatter_3D(X_data_ave_EM_plus[i][:, 0],
X_data_ave_EM_plus[i][:, 1],
X_data_ave_EM_plus[i][:, 2],
nf=0,
na=0)
gl.scatter_3D(0, 0, 0, nf=1, na=0)
for i in range(Nclasses):
gl.scatter_3D(X_data_ave_EM_minus[i][:, 0],
X_data_ave_EM_minus[i][:, 1],
X_data_ave_EM_minus[i][:, 2],
nf=0,
na=0)
## For the 2-EM across trials
gl.scatter_3D(0, 0, 0, nf=1, na=0)
for i in range(Nclasses):
gl.scatter_3D(X_trials_EM_plus[i][:, 0],
X_trials_EM_plus[i][:, 1],
X_trials_EM_plus[i][:, 2],
nf=0,
na=0)
gl.scatter_3D(0, 0, 0, nf=1, na=0)
for i in range(Nclasses):
gl.scatter_3D(X_trials_EM_minus[i][:, 0],
X_trials_EM_minus[i][:, 1],
X_trials_EM_minus[i][:, 2],
nf=0,
na=0)
import Watson_estimators as Wae
import general_func as gf
plt.close("all")
mu_caca = np.ones((1000, 1))
################################################################
######## Load and combine 3 sets ##############################
###############################################################
EM_data = 1
if (EM_data):
K = 3
#gl.scatter_3D([0,1,1,1,1,-1,-1,-1,-1], [0,1,1,-1,-1,1,1,-1,-1],[0,1,-1,1,-1,1,-1,1,-1], nf = 1, na = 0)
gl.scatter_3D(0, 0, 0, nf=1, na=0)
kflag = 0
for k in range(1, K + 1):
# folder = "./EM_data/"
folder = "./test_data/"
filedir = folder + "Wdata_" + str(k) + ".csv"
Xdata_k = np.array(pd.read_csv(filedir, sep=",", header=None))
Xdata_k = Xdata_k[:1000, :]
print Xdata_k.shape
# Xdata_param = pkl.load_pickle( folder + "Wdata_"+ str(k)+".pkl",1)
# mu = Xdata_param[0]
# kappa = Xdata_param[1]
# print "Real: ", mu,kappa
# Generate and plot the data
def get_2EM_vectors(X_All_labels, label_classes,
max_trials = 50, channel_sel= None,
plot_flag = 0):
Nclasses = len(X_All_labels)
# Instead of obtaining the average profile at each sime instant, we run
# a 2-EM clustering to each label, at each time instance.
# For each time instance,instead of getting the previous mean, we get 2 vectors.
# The mean vector of the class with high kappa and the one with low kappa
# We work on the assumtion that we have 2 clusters,
X_data_ave_EM_plus, X_data_ave_EM_minus = dp.get_labels_ave_EM(
X_All_labels, label_classes,
max_trials = max_trials, channel_sel= channel_sel)
# Get the 2-EM for all the trials !!
# For every trial we just run a 2 EM and get the direction of the
X_trials_EM_plus, X_trials_EM_minus = dp.get_X_trials_EM(
X_All_labels, label_classes,
max_trials = max_trials, channel_sel= channel_sel)
plot_flag = 0
if (plot_flag):
## For the 2-EM across time
gl.scatter_3D(0, 0,0, nf = 1, na = 0)
for i in range(Nclasses):
gl.scatter_3D(X_data_ave_EM_plus[i][:,0], X_data_ave_EM_plus[i][:,1],X_data_ave_EM_plus[i][:,2], nf = 0, na = 0)
gl.scatter_3D(0, 0,0, nf = 1, na = 0)
for i in range(Nclasses):
gl.scatter_3D(X_data_ave_EM_minus[i][:,0], X_data_ave_EM_minus[i][:,1],X_data_ave_EM_minus[i][:,2], nf = 0, na = 0)
## For the 2-EM across trials
gl.scatter_3D(0, 0,0, nf = 1, na = 0)
for i in range(Nclasses):
gl.scatter_3D(X_trials_EM_plus[i][:,0], X_trials_EM_plus[i][:,1],X_trials_EM_plus[i][:,2], nf = 0, na = 0)
gl.scatter_3D(0, 0,0, nf = 1, na = 0)
for i in range(Nclasses):
gl.scatter_3D(X_trials_EM_minus[i][:,0], X_trials_EM_minus[i][:,1],X_trials_EM_minus[i][:,2], nf = 0, na = 0)
# intervals etc.
# Fit the model
model = ols("Y ~ MACD + RSI + ATR + MACD_vel + ATR_vel + RSI_vel",
data).fit()
params = model._results.params
# Print the summary
print(model.summary())
print("OLS model Parameters")
print(params)
Xdata = np.array([MACD[:, indx], RSI[:, indx], MACD_vel[:, indx]]).T
## Graoca bonita !! TODO
print(ul.fnp(np.sqrt(np.sum(Xdata * Xdata, axis=1))).shape)
Xdata = Xdata / ul.fnp(np.sqrt(np.sum(Xdata * Xdata, axis=1)))
gl.scatter_3D(Xdata[:, 0], Xdata[:, 1], Xdata[:, 2])
gl.scatter_3D(-Xdata[:, 0], -Xdata[:, 1], -Xdata[:, 2], nf=0)
gl.scatter_3D(-Xdata[:, 0], Xdata[:, 1], -Xdata[:, 2], nf=0)
gl.scatter_3D(Xdata[:, 0], -Xdata[:, 1], Xdata[:, 2], nf=0)
# Peform analysis of variance on fitted linear model
#anova_results = anova_lm(model)
#print('\nANOVA results')
#print(anova_results)
##############################################################################
model_sklearn = 1
if (model_sklearn):
## Eliminate the Nans !! Even if they appear in just one dim
mask_X = np.sum(np.isnan(X_data), axis=1) == 0
gl.plot(lags, residuals)
############## Plotting of 3D regression ##############
plotting_3D = 1
if (plotting_3D):
# We train 3 models
# Y_data = np.sign(Y_data)
mask_X = np.sum(np.isnan(X_data), axis = 1) == 0
mask_Y = np.isnan(Y_data) == 0
mask = mask_X & mask_Y[:,0] # Mask without NaNs. This is done automatically in the OLS
# Create linear regression object
regr = linear_model.LinearRegression()
# Train the model using the training sets
X_datafiltered = X_data[:,[0,1]]
regr.fit(X_datafiltered[mask,:], Y_data[mask,:])
# coeffs = np.array([regr.intercept_, regr.coef_])[0]
coeffs = np.append(regr.intercept_, regr.coef_)
params = np.array(coeffs)
gl.scatter_3D(X_data[:,0],X_data[:,1], Y_data,
labels = ["","",""],
legend = ["Pene"],
nf = 1)
grids = ul.get_grids(X_data)
z = bMA.get_plane_Z(grids[0], grids[1],params)
h = gl.plot_3D(grids[0],grids[1],z, nf = 0)
# Own libraries
import import_folders
from graph_lib import gl
import sampler_lib as sl
import EM_lib as EMl
import pickle_lib as pkl
import Watson_distribution as Wad
import Watson_sampling as Was
import Watson_estimators as Wae
import general_func as gf
import copy
K = 3
gl.scatter_3D(0, 0, 0, nf=1, na=0)
N = 10000
Xdata = [] # List will all the generated data
K = 3
#gl.scatter_3D([0,1,1,1,1,-1,-1,-1,-1], [0,1,1,-1,-1,1,1,-1,-1],[0,1,-1,1,-1,1,-1,1,-1], nf = 1, na = 0)
gl.scatter_3D(0, 0, 0, nf=1, na=0)
kflag = 0
for k in range(1, K + 1):
folder = "./EM_data/"
folder = "./test_data/"
filedir = folder + "Wdata_" + str(k) + ".csv"
Xdata_k = np.array(pd.read_csv(filedir, sep=",", header=None))
Xdata_k = Xdata_k[:, :]
print Xdata_k.shape
Module = Module.reshape(Nsamples, 1)
# Check that the modulus is not 0
Xdata = Xdata[np.where(Module > tol)[0], :]
Xdata = np.divide(Xdata, Module)
gl.scatter(Xdata[:, 0], Xdata[:, 1], nf=0)
### POLLAS
Nsamples = 1000
Ndim = 3
tol = 0.000000001
Xdata = np.random.randn(Nsamples, Ndim) + 2
Module = np.sqrt(np.sum(np.power(Xdata, 2), 1))
Module = Module.reshape(Nsamples, 1)
# Check that the modulus is not 0
Xdata = Xdata[np.where(Module > tol)[0], :]
Xdata = np.divide(Xdata, Module)
gl.scatter_3D(Xdata[:, 0], Xdata[:, 1], Xdata[:, 2])
Nsamples = 100
Xdata = np.random.randn(Nsamples, Ndim) - 0
Module = np.sqrt(np.sum(np.power(Xdata, 2), 1))
Module = Module.reshape(Nsamples, 1)
# Check that the modulus is not 0
Xdata = Xdata[np.where(Module > tol)[0], :]
Xdata = np.divide(Xdata, Module)
gl.scatter_3D(Xdata[:, 0], Xdata[:, 1], Xdata[:, 2], nf=0)
folder_EM = "../data/EM_data/"
folder_HMM = "../data/HMM_data/"
generate_data = 1
if (generate_data):
### Generate the data, save it to file, load it and estimate parameters
kappa = 20
N = 1000
mu = np.array([2, 4, 5])
mu = mu / np.sqrt(np.sum(mu * mu))
RandWatson = Was.randWatson(N, mu, kappa)
gl.scatter_3D(RandWatson[:, 0],
RandWatson[:, 1],
RandWatson[:, 2],
nf=1,
na=0)
print "Generation Parameters"
print mu
print kappa
print "Estimation Parameters"
mu = Wae.get_MLmean(RandWatson)
kappa = Wae.get_MLkappa(mu, RandWatson)
print mu
print kappa
## Save the file to disk !!
filedir = folder_EM + "Wdata.csv"
folder = "./HMM_data/"
Chains_list = pkl.load_pickle(folder + "HMM_labels.pkl", 1)
HMM_list = pkl.load_pickle(folder + "HMM_datapoints.pkl", 1)
params = pkl.load_pickle(folder + "HMM_param.pkl", 1)
pi = params[0]
A = params[1]
pi_end = HMMlf.get_final_probabilities(pi, A, 20)
#print pi_end
print "Real pi"
print pi
print "Real A"
print A
gl.scatter_3D(0, 0, 0, nf=1, na=0)
for XdataChain in HMM_list:
gl.scatter_3D(XdataChain[:, 0],
XdataChain[:, 1],
XdataChain[:, 2],
nf=0,
na=0)
################################################################
######## Initialization of the parameters ##############################
###############################################################
I = 3
D = HMM_list[0].shape[1]
init_with_EM = 0
# Own libraries
import import_folders
from graph_lib import gl
import sampler_lib as sl
import EM_lib as EMl
import pickle_lib as pkl
import Watson_distribution as Wad
import Watson_sampling as Was
import Watson_estimators as Wae
import general_func as gf
import copy
K = 3
gl.scatter_3D(0, 0,0, nf = 1, na = 0)
N = 10000
Xdata = [] # List will all the generated data
K = 3
#gl.scatter_3D([0,1,1,1,1,-1,-1,-1,-1], [0,1,1,-1,-1,1,1,-1,-1],[0,1,-1,1,-1,1,-1,1,-1], nf = 1, na = 0)
gl.scatter_3D(0, 0,0, nf = 1, na = 0)
kflag = 0
for k in range(1,K+1):
folder = "./EM_data/"
folder = "./test_data/"
filedir = folder + "Wdata_"+ str(k)+".csv"
Xdata_k = np.array(pd.read_csv(filedir, sep = ",", header = None))
Xdata_k = Xdata_k[:,:]
print Xdata_k.shape
X_All_labels = [X_All_labels[0], X_All_labels[2]]
label_classes = [label_classes[0], label_classes[2]]
# Now X_All_labels has in every postion the trials for each class in the form
# of a matrix Ntrials x Ntimes x Ndim
################################################################
######## Preprocessing ! ##############################
###############################################################
# For the first label
plotting_all_trials_one_instant = 0
if (plotting_all_trials_one_instant):
gl.scatter_3D(0, 0, 0, nf=1, na=0)
Ntrial, Nsam, Ndim = X_All_labels[0].shape
i0 = 50 # Good one !!
t_show = 30
for i in range(Ntrial):
X_trial = X_All_labels[0][i, :, i0:i0 + 3] # Nsam x Ndim
# X_trial = X_trial - np.sum(X_trial, axis = 1).reshape(X_trial.shape[0],1)
# scaler = preprocessing.StandardScaler().fit(X_trial)
# X_trial = scaler.transform(X_trial)
X_trial = gf.normalize_data(X_trial)
gl.scatter_3D(X_trial[t_show, 0],
X_trial[t_show, 1],
X_trial[t_show, 2],
for i in range(Nsa):
probs.append([])
for j in range(Nsa):
XdataSample = [np.sin(Xthetta[i])*np.cos(Xfi[j]),
np.sin(Xthetta[i])*np.sin(Xfi[j]),
np.cos(Xthetta[i])]
probs[i].append(Wad.Watson_pdf(XdataSample,mu,kappa ))
probs = np.array(probs).T
## Plotting
gl.set_subplots(1,3)
## Plot it in terms of (angle, prob)
gl.plot_3D(Xthetta,Xfi, np.array(probs))
gl.plot_3D(Xthetta,Xfi, np.array(probs), project = "spher")
mu = np.random.randn(5,1);
mu = gf.normalize_module(mu.T).flatten()
## Generate samples
RandWatson = Was.randWatson(Nsampling, mu, kappa)
gl.scatter_3D(RandWatson[:,0],RandWatson[:,1], RandWatson[:,2])
mu_est = Wae.get_MLmean(RandWatson)
kappa_est = Wae.get_MLkappa(mu_est, RandWatson)
mu_est2, kappa_est2 = Wae.get_Watson_muKappa_ML(RandWatson)
print "Real: ", mu, kappa
print "Estimate: ", mu_est2, kappa_est2
plt.close("all")
folder = "./data/test_data/"
folder_HMM = "./data/HMM_data/"
load_EM_data = 0
load_HMM_data = 1
perform_EM = 1
################################################################
######## Load and combine 3 sets ##############################
###############################################################
if (load_EM_data):
K = 3
#gl.scatter_3D([0,1,1,1,1,-1,-1,-1,-1], [0,1,1,-1,-1,1,1,-1,-1],[0,1,-1,1,-1,1,-1,1,-1], nf = 1, na = 0)
gl.scatter_3D(0, 0, 0, nf=1, na=0)
kflag = 0
for k in range(1, K + 1):
filedir = folder + "Wdata_" + str(k) + ".csv"
Xdata_k = np.array(pd.read_csv(filedir, sep=",", header=None))
Xdata_k = Xdata_k[:1000, :]
# Xdata_param = pkl.load_pickle( folder + "Wdata_"+ str(k)+".pkl",1)
# mu = Xdata_param[0]
# kappa = Xdata_param[1]
# print "Real: ", mu,kappa
# Generate and plot the data
gl.scatter_3D(Xdata_k[:, 0], Xdata_k[:, 1], Xdata_k[:, 2], nf=0, na=0)
