def test_mysvc_reducer(self):
## 1) Build dataset
## ===================================================================
X, y = datasets.make_classification(n_samples=12,
n_features=10,
n_informative=2,
random_state=1)
## 2) run with Methods
## ===================================================================
my_svc1 = MySVC(C=1.0)
my_svc2 = MySVC(C=2.0)
two_svc_single = Methods(my_svc1, my_svc2)
two_svc_local = Methods(my_svc1, my_svc2)
two_svc_swf = Methods(my_svc1, my_svc2)
two_svc_single.reducer = MyReducer()
two_svc_local.reducer = MyReducer()
two_svc_swf.reducer = MyReducer()
for leaf in two_svc_single.walk_leaves():
print leaf.get_key()
for leaf in two_svc_local.walk_leaves():
print leaf.get_key()
for leaf in two_svc_swf.walk_leaves():
print leaf.get_key()
# top-down process to call transform
two_svc_single.run(X=X, y=y)
# buttom-up process to compute scores
res_single = two_svc_single.reduce()
### You can get below results:
### ==================================================================
### [{'MySVC(C=1.0)': array([ 1., 1.])}, {'MySVC(C=2.0)': array([ 1., 1.])}]
### 3) Run using local multi-processes
### ==================================================================
from epac.map_reduce.engine import LocalEngine
local_engine = LocalEngine(two_svc_local, num_processes=2)
two_svc_local = local_engine.run(**dict(X=X, y=y))
res_local = two_svc_local.reduce()
### 4) Run using soma-workflow
### ==================================================================
from epac.map_reduce.engine import SomaWorkflowEngine
sfw_engine = SomaWorkflowEngine(tree_root=two_svc_swf,
num_processes=2)
two_svc_swf = sfw_engine.run(**dict(X=X, y=y))
res_swf = two_svc_swf.reduce()
if not repr(res_swf) == repr(res_local):
raise ValueError("Cannot dump class definition")
if not repr(res_swf) == repr(res_single):
raise ValueError("Cannot dump class definition")
def test_mysvc_reducer(self):
## 1) Build dataset
## ===================================================================
X, y = datasets.make_classification(n_samples=12,
n_features=10,
n_informative=2,
random_state=1)
## 2) run with Methods
## ===================================================================
my_svc1 = MySVC(C=1.0)
my_svc2 = MySVC(C=2.0)
two_svc_single = Methods(my_svc1, my_svc2)
two_svc_local = Methods(my_svc1, my_svc2)
two_svc_swf = Methods(my_svc1, my_svc2)
two_svc_single.reducer = MyReducer()
two_svc_local.reducer = MyReducer()
two_svc_swf.reducer = MyReducer()
for leaf in two_svc_single.walk_leaves():
print(leaf.get_key())
for leaf in two_svc_local.walk_leaves():
print(leaf.get_key())
for leaf in two_svc_swf.walk_leaves():
print(leaf.get_key())
# top-down process to call transform
two_svc_single.run(X=X, y=y)
# buttom-up process to compute scores
res_single = two_svc_single.reduce()
### You can get below results:
### ==================================================================
### [{'MySVC(C=1.0)': array([ 1., 1.])}, {'MySVC(C=2.0)': array([ 1., 1.])}]
### 3) Run using local multi-processes
### ==================================================================
from epac.map_reduce.engine import LocalEngine
local_engine = LocalEngine(two_svc_local, num_processes=2)
two_svc_local = local_engine.run(**dict(X=X, y=y))
res_local = two_svc_local.reduce()
### 4) Run using soma-workflow
### ==================================================================
from epac.map_reduce.engine import SomaWorkflowEngine
sfw_engine = SomaWorkflowEngine(tree_root=two_svc_swf, num_processes=2)
two_svc_swf = sfw_engine.run(**dict(X=X, y=y))
res_swf = two_svc_swf.reduce()
if not repr(res_swf) == repr(res_local):
raise ValueError("Cannot dump class definition")
if not repr(res_swf) == repr(res_single):
raise ValueError("Cannot dump class definition")
def test_constructor_avoid_collision_level2(self):
# Test that level 2 collisions are avoided
pm = Methods(*[Pipe(SelectKBest(k=2), SVC(kernel="linear", C=C))\
for C in [1, 10]])
leaves_key = [l.get_key() for l in pm.walk_leaves()]
self.assertTrue(len(leaves_key) == len(set(leaves_key)),
u'Collision could not be avoided')
def test_constructor_avoid_collision_level2(self):
# Test that level 2 collisions are avoided
pm = Methods(*[Pipe(SelectKBest(k=2), SVC(kernel="linear", C=C))
for C in [1, 10]])
leaves_key = [l.get_key() for l in pm.walk_leaves()]
self.assertTrue(len(leaves_key) == len(set(leaves_key)),
u'Collision could not be avoided')
def test_constructor_avoid_collision_level1(self):
# Test that level 1 collisions are avoided
pm = Methods(*[SVC(kernel="linear", C=C) for C in [1, 10]])
leaves_key = [l.get_key() for l in pm.walk_leaves()]
self.assertTrue(len(leaves_key) == len(set(leaves_key)), u"Collision could not be avoided")
wf = CVBestSearchRefit(Methods(SVM(C=1), SVM(C=10)))
wf.run(X=X, y=y)
print(wf.reduce())
# Feature selection combined with SVM and LDA
# CVBestSearchRefit
# Methods (Splitter)
# / \
# KBest(1) KBest(5) SelectKBest (Estimator)
# |
# Methods (Splitter)
# / \
# LDA() SVM() ... Classifiers (Estimator)
pipelines = Methods(
*[Pipe(SelectKBest(k=k), Methods(LDA(), SVM())) for k in [1, 5]])
print([n for n in pipelines.walk_leaves()])
best_cv = CVBestSearchRefit(pipelines)
best_cv.run(X=X, y=y)
best_cv.reduce()
# Put it in an outer CV
cv = CV(best_cv)
cv.run(X=X, y=y)
cv.reduce()
# Perms + Cross-validation of SVM(linear) and SVM(rbf)
# -------------------------------------
# Perms Perm (Splitter)
# / | \
# 0 1 2 Samples (Slicer)
# |
# CV + Grid search of a simple classifier wf = CVBestSearchRefit(Methods(SVM(C=1), SVM(C=10))) wf.run(X=X, y=y) print wf.reduce() # Feature selection combined with SVM and LDA # CVBestSearchRefit # Methods (Splitter) # / \ # KBest(1) KBest(5) SelectKBest (Estimator) # | # Methods (Splitter) # / \ # LDA() SVM() ... Classifiers (Estimator) pipelines = Methods(*[Pipe(SelectKBest(k=k), Methods(LDA(), SVM())) for k in [1, 5]]) print [n for n in pipelines.walk_leaves()] best_cv = CVBestSearchRefit(pipelines) best_cv.run(X=X, y=y) best_cv.reduce() # Put it in an outer CV cv = CV(best_cv) cv.run(X=X, y=y) cv.reduce() # Perms + Cross-validation of SVM(linear) and SVM(rbf) # ------------------------------------- # Perms Perm (Splitter) # / | \ # 0 1 2 Samples (Slicer) # |
print svms_auto.refited.estimator.coef_
##############################################################################
# Put everything together
# Pipeline, "Pipe": SelectKBest + StandardScaler + SVM l1 vs l2
from epac import range_log2
from epac import CVBestSearchRefit, Pipe, Methods, CV
k_values = range_log2(X.shape[1], add_n=True)
C_values = [.1, 1, 10, 100]
anova_svms = Methods(*[Pipe(SelectKBest(k=k), preprocessing.StandardScaler(),
Methods(*[SVM(C=C, penalty=penalty, class_weight='auto', dual=False)
for C in C_values for penalty in ['l1', 'l2']]))
for k in k_values])
# Take a look
print [l for l in anova_svms.walk_leaves()]
## k and C selection based on CV
anova_svms_auto = CVBestSearchRefit(anova_svms)
#anova_svm_all = Methods(anova_svm, anova_svm_cv)
cv = CV(anova_svms_auto, n_folds=n_folds)
time_fit_predict = time.time()
cv.run(X=X, y=y)
print time.time() - time_fit_predict
print cv.reduce()
# Re-fit on all data. Warning: biased !!!
anova_svms_auto.run(X=X, y=y)
print anova_svms_auto.best_params
