def plot_NoteAndSens():
# Plot time series of note (volume) and sensors of specific file
# TODO: save it to the folder
# files = next(os.walk(directory))[1]
# for f in files:
f = 'two_scales_1'
A = Analyze(filename=f)
A.plotKeyAndGlove('basic', savename=f)
def fit_sparse(self, feature_opt='1storder', variableCoef=False, variableCoefOrder=0, variableCoefBasis='simple_polynomial', \
LassoType='Lasso', RegCoef=0.00001, cv=None, criterion=None, maxiter=10000, tolerance=0.00001, use_rfe=False, normalize=True,
rfe_iter=10, rfe_alpha=None, print_rfeiter=False, shuffle=False, nzthresh=1e-200, basefile='', adjustgrid={}, save=True,
comments='', checkExistence=True):
# Make Metadata and Check its existence
metadata = self.makeMetadata(basefile, adjustgrid, feature_opt, self.trainratio, variableCoef, variableCoefOrder, variableCoefBasis, \
LassoType, cv, criterion, use_rfe, rfe_alpha, nzthresh, maxiter, comments)
datahandler = DataIO(self.scase, directory=LEARNDIR, basefile=basefile)
if checkExistence:
exists, filename = datahandler.checkMetadataInDir(
metadata, ignore_prop='nzthresh')
if exists:
return filename + '.txt'
# Make features and training set
F = Features(scase=self.scase,
option=feature_opt,
variableCoef=variableCoef,
variableCoefOrder=variableCoefOrder,
variableCoefBasis=variableCoefBasis)
self.featurelist, self.labels, self.featurenames = F.makeFeatures(
self.grid, self.fu, self.ICparams)
Xtrain, ytrain, Xtest, ytest = self.makeTTsets(self.featurelist,
self.labels,
shuffle=shuffle,
threshold=nzthresh)
# Choose optimization algorithm
lin = self.choose_optimizer(LassoType=LassoType,
RegCoef=RegCoef,
cv=cv,
criterion=criterion,
maxiter=maxiter,
tolerance=tolerance,
normalize=normalize)
# Train model with Lasso
# Choose to use Recursive Feature Elimination or not
if use_rfe:
lin, rem_feature_idx = self.train_rfe(lin,
Xtrain,
ytrain,
rfe_iter=rfe_iter,
rfe_alpha=rfe_alpha,
print_rfeiter=print_rfeiter)
Xtrain = Xtrain[:, rem_feature_idx]
Xtest = Xtest[:, rem_feature_idx]
coefficients = lin.coef_
else:
lin, rem_feature_idx = self.train_single(lin, Xtrain, ytrain)
coefficients = lin.coef_[rem_feature_idx]
# Outputs
output = {}
# Compute Erros and Scores
output['trainRMSE'] = np.sqrt(
mean_squared_error(ytrain, lin.predict(Xtrain)))
output['testRMSE'] = np.sqrt(
mean_squared_error(ytest, lin.predict(Xtest)))
output['trainScore'] = lin.score(Xtrain, ytrain)
output['testScore'] = lin.score(Xtest, ytest)
rem_featurenames = [self.featurenames[i] for i in rem_feature_idx]
output['featurenames'] = rem_featurenames
output['coef'] = coefficients.tolist() # Might not work for RFE !!
output['n_iter'] = lin.n_iter_
# Different optimizers have different outputs
if LassoType == 'LassoLarsIC':
output['alpha'] = lin.alpha_.tolist()
output['criterion_path'] = lin.criterion_.tolist()
elif LassoType == 'LassoCV':
output['alpha'] = lin.alpha_.tolist()
output['alpha_mse_path'] = lin.mse_path_.mean(axis=1).tolist()
output['alpha_path'] = lin.alphas_.tolist()
output['dual_gap'] = lin.dual_gap_
elif LassoType in {'LassoLarsCV', 'LarsCV'}:
output['alpha'] = lin.alpha_
output['alpha_mse_path'] = lin.mse_path_.mean(
axis=1).tolist() # Average along CV folds
output['cv_alpha_path'] = lin.cv_alphas_.tolist(
) # Goes with mse_path
output['coef_path'] = lin.coef_path_.tolist()
output['alpha_path'] = lin.alphas_.tolist() # Goes with coef_path
elif LassoType == 'Lasso':
output['alpha'] = RegCoef
# Printing
if self.verbose:
A = Analyze()
A.print_results(output, metadata)
# Saving
filename = datahandler.saveSolution(output,
metadata,
fileformat='.txt')
return filename
# Save Learning
D = DataIO(case, directory=LEARNDIR)
output, metadata = D.readLearningResults(filename)
output_vec.append(output)
metadata_vec.append(metadata)
filename_vec.append(filename)
print('files = [')
for f in filename_vec:
print("\'"+f+"\',")
print(']')
## PLOT
A = Analyze()
savename = 'Burgers_MC_CDF'
A.plotRMSEandCoefs(output_vec, MCcountvec, '$N_{MC}$, Number of Realizations', threshold=0.01, set_grid=False, cdf=True, invert_sign=True, savename='Burgers_MC_CDF')
##############
##Function of Regularization - For MC = num_realizations (MAX)
## WTF??
# fig, ax = plt.subplots(1, 2)
# alphas, mse = A.getRegMseDependence_single(output_vec[-1])
# ax[0].plot(alphas, mse)
# ax[0].set_xlabel('Regularization Coefficient')
# ax[0].set_ylabel('MSE')
# alphas, coefficients, feats = A.getCoefRegDependence(output_vec[-1], threshold=0.0)
# for i in range(len(feats)):
output_vec.append(output)
metadata_vec.append(metadata)
filename_vec.append(filename)
output_VEC.append(output_vec)
metadata_VEC.append(metadata_vec)
filename_VEC.append(filename_vec)
print('files = [')
for f in filename_vec:
print("\'" + f + "\',")
print(']')
## PLOT
A = Analyze()
savename = 'advectreact_rfe' + "_closecomp_" + LassoType + "_" + str(coeforder)
xlabel = 'RFE Threshold'
feature_opt_vec = ['Closure Prob.', 'Full Eqn.']
linestyles = ['solid', 'dashed', 'dashdot', 'dotted']
marker = ['o', 'v', 's',
'*'] #, '^', '>', '<', 'x', 'D', '1', '.', '2', '3', '4']
styles = [[l, m] for l in linestyles for m in marker]
print(styles)
print(len(styles))
fig, ax = plt.subplots(1, 2, figsize=(13, 5.5))
variable = rfe_alpha_vec
leg0 = []
def plotRun(self):
dataplotting = Analyze(self.directory, self.filename)
dataplotting.plotKeyAndGlove('basic')
dataplotting.plot_sampling_frequency()
metadata_vec.append(metadata)
filename_vec.append(filename)
output_VEC.append(output_vec)
metadata_VEC.append(metadata_vec)
filename_VEC.append(filename_vec)
print('files = [')
for f in filename_vec:
print("\'"+f+"\',")
print(']')
## PLOT
A = Analyze()
savename = 'advectreact_rfe' + "_ordercomp_" + LassoType + "_" + feature_opt
xlabel = 'RFE Threshold'
linestyles = ['solid', 'dashed', 'dashdot', 'dotted']
marker = ['o', 'v', 's', '*']#, '^', '>', '<', 'x', 'D', '1', '.', '2', '3', '4']
styles = [[l, m] for l in linestyles for m in marker]
fig, ax = plt.subplots(1, 2, figsize=(13, 5.5))
variable = rfe_alpha_vec
leg0 = []
leg1 = []
for i, outvec in enumerate(output_VEC):
trainRMSE, testRMSE = A.getTrainTestDependence(outvec)
# READ Learning
D = DataIO(case, directory=LEARNDIR)
output, metadata = D.readLearningResults(filename)
output_vec.append(output)
metadata_vec.append(metadata)
filename_vec.append(filename)
print('files = [')
for f in filename_vec:
print("\'" + f + "\',")
print(']')
## PLOT
A = Analyze()
portion_from_boundary = [m[0] for m in muvec]
savename = 'advectreact_boundary'
# ax = A.plotRMSEandCoefs(output_vec, portion_from_boundary, 'Distance from $U=0$ (Domain Fraction)', \
# set_grid=False, cdf=False, threshold=0.01, invert_sign=True, savename=savename)
#####
variable = portion_from_boundary
xlabel = 'Distance from $U=0$ (Domain Fraction)'
fig, ax = plt.subplots(1, 2, figsize=(9, 4))
# Coefficients Dependence Multi
featarray, relevant_feats = A.getCoefDependence(output_vec,
threshold=0.01,
# READ Learning
D = DataIO(case, directory=LEARNDIR)
output, metadata = D.readLearningResults(filename)
output_vec.append(output)
metadata_vec.append(metadata)
filename_vec.append(filename)
print('files = [')
for f in filename_vec:
print("\'" + f + "\',")
print(']')
## PLOT
A = Analyze()
savename = 'advectreact_rde'
A.plotRMSEandCoefs(output_vec,
rfe_alpha_vec,
'Percentage Distance from Boundary',
threshold=0.01,
invert_sign=True,
savename=savename)
# Plot boundary
s = 8
V = Visualize(grid)
V.plot_fu3D(fu)
V.plot_fu(fu, dim='t', steps=s)
V.plot_fu(fu, dim='x', steps=s)
V.show()
# READ Learning
D = DataIO(case, directory=LEARNDIR)
output, metadata = D.readLearningResults(filename)
output_vec.append(output)
metadata_vec.append(metadata)
filename_vec.append(filename)
print('files = [')
for f in filename_vec:
print("\'" + f + "\',")
print(']')
## PLOT
A = Analyze()
savename = 'advectreact_MC' + "_" + LassoType + "_" + str(mu[0]).split('.')[1]
A.plotRMSEandCoefs(output_vec,
MCcountvec,
'Number of Realizations',
threshold=0.01,
invert_sign=True,
use_logx=False,
set_grid=True,
savename=savename)
## PLOT 3D
# s = 8
# V = Visualize(grid)
# V.plot_fu3D(fu)
scase=case,
trainratio=trainratio,
verbose=True)
filename = difflearn.fit_sparse(feature_opt=feature_opt, variableCoef=variableCoef, variableCoefBasis=variableCoefBasis, \
variableCoefOrder=coeforder, use_rfe=use_rfe, rfe_alpha=rfe_alpha, nzthresh=nzthresh, maxiter=maxiter, \
LassoType=LassoType, RegCoef=RegCoef, cv=cv, criterion=criterion, print_rfeiter=print_rfeiter, shuffle=shuffle, \
basefile=savenamepdf, adjustgrid=adjustgrid, save=save, normalize=normalize, comments=comments)
# READ Learning
D = DataIO(case, directory=LEARNDIR)
output, metadata = D.readLearningResults(filename)
output_vec.append(output)
metadata_vec.append(metadata)
filename_vec.append(filename)
print('files = [')
for f in filename_vec:
print("\'" + f + "\',")
print(']')
## PLOT
A = Analyze()
savename = 'advectreact_randadv_closure'
A.barRMSEandCoefs(output_vec,
feature_opt_vec,
'Method',
threshold=0.01,
invert_sign=True,
savename=savename)
trainratio=trainratio,
verbose=printlearning)
filename = difflearn.fit_sparse(feature_opt=feature_opt, variableCoef=variableCoef, variableCoefBasis=variableCoefBasis, \
variableCoefOrder=coeforder, use_rfe=use_rfe, rfe_alpha=rfe_alpha, nzthresh=nzthresh, maxiter=maxiter, \
LassoType=LassoType, RegCoef=RegCoef, cv=cv, criterion=criterion, print_rfeiter=print_rfeiter, shuffle=shuffle, \
basefile=savenamepdf, adjustgrid=adjustgrid, save=save, normalize=normalize, comments=comments)
# READ Learning
D = DataIO(case, directory=LEARNDIR)
output, metadata = D.readLearningResults(filename)
output_vec.append(output)
metadata_vec.append(metadata)
filename_vec.append(filename)
## PLOT
# Error function of MC
A = Analyze()
savename = 'Burgers_rfe'
A.plotRMSEandCoefs(output_vec,
rfe_alpha_vec,
'RFE Threshold',
threshold=0.01,
invert_sign=True,
savename=savename)
plt.show()
# Plot Coefficients as a function of t0
basefile=savenamepdf, adjustgrid=adjustgrid, save=save, normalize=normalize, comments=comments)
# READ Learning
D = DataIO(case, directory=LEARNDIR)
output, metadata = D.readLearningResults(filename)
output_vec.append(output)
metadata_vec.append(metadata)
filename_vec.append(filename)
## PLOT
# Error function of MC
# fig = plt.figure()
A = Analyze()
savename = 'Burgers_shock'
portion = [(t[1] - 0.5) / (t[1] - t[0]) for t in mtvec]
A.plotRMSEandCoefs(output_vec,
portion,
'$p_s$, Time Portion in Shock Region',
threshold=0.01,
invert_sign=True,
cdf=True,
set_grid=False,
savename='Burgers_shock')
# trainRMSE, testRMSE = A.getTrainTestDependence(output_vec)
# t0 = [t[0] for t in mtvec]
# plt.plot(t0, testRMSE, linwidth=3)
# READ Learning
D = DataIO(case, directory=LEARNDIR)
output, metadata = D.readLearningResults(filename)
output_vec.append(output)
metadata_vec.append(metadata)
filename_vec.append(filename)
print('files = [')
for f in filename_vec:
print("\'" + f + "\',")
print(']')
## PLOT
A = Analyze()
savename = 'advectreact_rfe' + "_" + feature_opt + "_" + LassoType + "_" + str(
coeforder)
A.plotRMSEandCoefs(output_vec,
rfe_alpha_vec,
'RFE Threshold',
threshold=0.001,
use_logx=True,
set_grid=True,
invert_sign=True,
savename=savename,
show=True)
except:
print("\n\n\n************************\n\n\n")
print("Exception Happened for ", feature_opt, " ", LassoType,
