Skip to content

Commit

Permalink
Merge branch 'devel'
Browse files Browse the repository at this point in the history
  • Loading branch information
@Morisset
Morisset committed Sep 27, 2019
2 parents 3fb5a7d + 059f3ea commit 8e8eddc
Show file tree
Hide file tree
Showing 5 changed files with 846 additions and 79 deletions.
3 changes: 3 additions & 0 deletions .gitignore
View file
Original file line number Diff line number Diff line change
Expand Up @@ -54,3 +54,6 @@ nosetests.xml
coverage.xml
*.cover

*.mwinai_sk
*.mwinai_k0
*.mwinai_k1
82 changes: 62 additions & 20 deletions docs/ComparePolynom.ipynb
View file

Large diffs are not rendered by default.

686 changes: 686 additions & 0 deletions docs/SaveRestore.ipynb
View file

Large diffs are not rendered by default.

152 changes: 94 additions & 58 deletions mwinai/Regressor/RegressionModel.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,7 @@
import numpy as np
import time
import random
from glob import glob
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler, RobustScaler
from sklearn.externals import joblib
Expand Down Expand Up @@ -65,7 +66,8 @@ def __init__(self, RM_type = 'SK_ANN',
use_log=False,
split_ratio = None,
verbose=False,
RM_filename=None, use_RobustScaler=False,
RM_filename=None,
use_RobustScaler=False,
random_seed=None,
noise=None,
N_y_bins=None,
Expand Down Expand Up @@ -630,14 +632,19 @@ def predict(self, scoring=True, reduce_by=None):
if self.N_test != self.N_test_y:
raise Exception('N_test {} != N_test_y {}'.format(self.N_test, self.N_test_y))
if self._multi_predic:
self.predic_score = [RM.score(self.X_test, self.y_test) for RM in self.RMs]
try:
self.predic_score = [RM.score(self.X_test, self.y_test) for RM in self.RMs]
except:
self.predic_score = [np.nan for RM in self.RMs]
else:
if self.y_train.ndim == 1:
y_tests = (self.y_test,)
else:
y_tests = self.y_test.T

self.predic_score = [RM.score(self.X_test, y_test) for RM, y_test in zip(self.RMs, y_tests)]
try:
self.predic_score = [RM.score(self.X_test, y_test) for RM, y_test in zip(self.RMs, y_tests)]
except:
self.predic_score = [np.nan for RM, y_test in zip(self.RMs, y_tests)]
if self.N_out != self.N_out_test:
raise Exception('N_out {} != N_out_test {}'.format(self.N_out,
self.N_out_test))
Expand Down Expand Up @@ -713,15 +720,23 @@ def save_RM(self, filename='RM', save_train=False, save_test=False):
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time, self.random_seed,
self.noise, X_train, y_train, X_test, y_test,
self.train_scaled, self.test_scaled, self.verbose), filename+'.mwinai1')
self.train_scaled, self.test_scaled, self.verbose), filename+'.mwinai_sk')
elif self.RM_type[0:2] == 'K_':
pass

joblib.dump((self.RM_type, self.RM_version,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time, self.random_seed,
self.noise, X_train, y_train, X_test, y_test,
self.train_scaled, self.test_scaled, self.verbose), filename+'.mwinai_k0')
for i, RM in enumerate(self.RMs):
RM.save('{}.mwinai_k{}'.format(filename, i+1))

if self.verbose:
print('RM save to {}'.format(filename))


def load_RM(self, filename='RM_jl.sav'):
def load_RM(self, filename='RM'):
"""
Loading previously saved model.
joblib is used to load.
Expand All @@ -739,56 +754,77 @@ def load_RM(self, filename='RM_jl.sav'):
RM.predict(scoring=False)
"""

RM_tuple = joblib.load(filename)
if self.RM_version != RM_tuple[1] and self.verbose:
print('WARNING: version loaded from {} is {}. Version from RM class is {}.'.format(filename,
RM_tuple[1],self.RM_version))
if RM_tuple[1] in ("0.15"):
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time, self.random_seed,
self.noise, self.X_train, self.y_train, self.X_test, self.y_test,
self.train_scaled, self.test_scaled, self.verbose) = RM_tuple
elif RM_tuple[1] in ("0.14"):
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time, self.random_seed, self.noise) = RM_tuple
elif RM_tuple[1] in ("0.12", "0.13"):
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time, self.random_seed) = RM_tuple
elif RM_tuple[1] == "0.11":
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time) = RM_tuple
elif RM_tuple[1] == "0.10":
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca) = RM_tuple
elif "{:.1f}".format(RM_tuple[1]) == "0.9":
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y) = RM_tuple
elif "{:.1f}".format(RM_tuple[1]) == "0.8":
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y) = RM_tuple
self.trained = True
if self.verbose:
print('RM loaded from {}'.format(filename))
files = glob("{}.*".format(filename))
if "{}.mwinai_sk".format(filename) in files:
RM_tuple = joblib.load("{}.mwinai_sk".format(filename))
if self.RM_version != RM_tuple[1] and self.verbose:
print('WARNING: version loaded from {} is {}. Version from RM class is {}.'.format(filename,
RM_tuple[1],self.RM_version))
if RM_tuple[1] in ("0.15"):
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time, self.random_seed,
self.noise, self.X_train, self.y_train, self.X_test, self.y_test,
self.train_scaled, self.test_scaled, self.verbose) = RM_tuple
elif RM_tuple[1] in ("0.14"):
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time, self.random_seed, self.noise) = RM_tuple
elif RM_tuple[1] in ("0.12", "0.13"):
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time, self.random_seed) = RM_tuple
elif RM_tuple[1] == "0.11":
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time) = RM_tuple
elif RM_tuple[1] == "0.10":
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca) = RM_tuple
elif "{:.1f}".format(RM_tuple[1]) == "0.9":
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y) = RM_tuple
elif "{:.1f}".format(RM_tuple[1]) == "0.8":
(self.RM_type, self.RM_version, self.RMs,
self.scaler, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y) = RM_tuple
self.trained = True
if self.verbose:
print('RM loaded from {}.mwinai_sk'.format(filename))

elif "{}.mwinai_k0".format(filename) in files:
RM_tuple = joblib.load("{}.mwinai_k0".format(filename))
if self.RM_version != RM_tuple[1] and self.verbose:
print('WARNING: version loaded from {} is {}. Version from RM class is {}.'.format(filename,
RM_tuple[1],self.RM_version))
if RM_tuple[1] in ("0.15"):
(self.RM_type, self.RM_version,
self.scaler, self.scaler_y, self.scaling_y, self.train_score, self._multi_predic,
self.N_in, self.N_out, self.N_in_test, self.N_out_test,
self.N_test, self.N_test_y, self.N_train, self.N_train_y,
self.pca_N, self.pca, self.training_time, self.random_seed,
self.noise, self.X_train, self.y_train, self.X_test, self.y_test,
self.train_scaled, self.test_scaled, self.verbose) = RM_tuple
if self.verbose:
print('RM loaded from {}.mwinai_k0'.format(filename))
self.RMs = [load_model("{}.mwinai_k1".format(filename))]
self.trained = True
if self.verbose:
print('RM loaded from {}.mwinai_k1'.format(filename))

#%%
if __name__ == "__main__":
Expand Down
2 changes: 1 addition & 1 deletion mwinai/version.py
View file
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
# -*- coding: utf-8 -*-
# mwinai version

__version__="0.2.1"
__version__="0.2.2"

0 comments on commit 8e8eddc

Please sign in to comment.