From 43456222fd2db3ad509986255e3751abdb5fd15d Mon Sep 17 00:00:00 2001
From: Baptiste Bauvin <baptiste.bauvin@lis-lab.fr>
Date: Thu, 20 Feb 2020 13:53:37 +0100
Subject: [PATCH] added tracebacks tracking in HPS
---
.../utils/hyper_parameter_search.py | 62 ++++++++++++-------
1 file changed, 39 insertions(+), 23 deletions(-)
diff --git a/multiview_platform/mono_multi_view_classifiers/utils/hyper_parameter_search.py b/multiview_platform/mono_multi_view_classifiers/utils/hyper_parameter_search.py
index df22a183..7d6b2ca4 100644
--- a/multiview_platform/mono_multi_view_classifiers/utils/hyper_parameter_search.py
+++ b/multiview_platform/mono_multi_view_classifiers/utils/hyper_parameter_search.py
@@ -3,6 +3,7 @@ import sys
import matplotlib.pyplot as plt
import numpy as np
+import traceback
from scipy.stats import randint, uniform
from sklearn.model_selection import RandomizedSearchCV
@@ -181,7 +182,8 @@ class MultiviewCompatibleRandomizedSearchCV(RandomizedSearchCV):
elif self.framework == "multiview":
return self.fit_multiview(X, y=y, groups=groups,**fit_params)
- def fit_multiview(self, X, y=None, groups=None, **fit_params):
+ def fit_multiview(self, X, y=None, groups=None, track_tracebacks=True,
+ **fit_params):
n_splits = self.cv.get_n_splits(self.available_indices, y[self.available_indices])
folds = list(self.cv.split(self.available_indices, y[self.available_indices]))
if self.equivalent_draws:
@@ -195,30 +197,44 @@ class MultiviewCompatibleRandomizedSearchCV(RandomizedSearchCV):
results = {}
self.cv_results_ = dict(("param_"+param_name, []) for param_name in candidate_params[0].keys())
self.cv_results_["mean_test_score"] = []
+ n_failed = 0
+ tracebacks = []
for candidate_param_idx, candidate_param in enumerate(candidate_params):
test_scores = np.zeros(n_splits)+1000
- for fold_idx, (train_indices, test_indices) in enumerate(folds):
- current_estimator = clone(base_estimator)
- current_estimator.set_params(**candidate_param)
- current_estimator.fit(X, y,
- train_indices=self.available_indices[train_indices],
- view_indices=self.view_indices)
- test_prediction = current_estimator.predict(
- X,
- self.available_indices[test_indices],
- view_indices=self.view_indices)
- test_score = self.scoring._score_func(y[self.available_indices[test_indices]],
- test_prediction,
- **self.scoring._kwargs)
- test_scores[fold_idx] = test_score
- for param_name, param in candidate_param.items():
- self.cv_results_["param_"+param_name].append(param)
- cross_validation_score = np.mean(test_scores)
- self.cv_results_["mean_test_score"].append(cross_validation_score)
- results[candidate_param_idx] = cross_validation_score
- if cross_validation_score <= min(results.values()):
- self.best_params_ = candidate_params[candidate_param_idx]
- self.best_score_ = cross_validation_score
+ try:
+ for fold_idx, (train_indices, test_indices) in enumerate(folds):
+ current_estimator = clone(base_estimator)
+ current_estimator.set_params(**candidate_param)
+ current_estimator.fit(X, y,
+ train_indices=self.available_indices[train_indices],
+ view_indices=self.view_indices)
+ test_prediction = current_estimator.predict(
+ X,
+ self.available_indices[test_indices],
+ view_indices=self.view_indices)
+ test_score = self.scoring._score_func(y[self.available_indices[test_indices]],
+ test_prediction,
+ **self.scoring._kwargs)
+ test_scores[fold_idx] = test_score
+ for param_name, param in candidate_param.items():
+ self.cv_results_["param_"+param_name].append(param)
+ cross_validation_score = np.mean(test_scores)
+ self.cv_results_["mean_test_score"].append(cross_validation_score)
+ results[candidate_param_idx] = cross_validation_score
+ if cross_validation_score <= min(results.values()):
+ self.best_params_ = candidate_params[candidate_param_idx]
+ self.best_score_ = cross_validation_score
+ except:
+ if track_tracebacks:
+ n_failed += 1
+ tracebacks.append(traceback.format_exc())
+ else:
+ raise
+ if n_failed == self.n_iter:
+ raise ValueError(
+ 'No fits were performed. All HP combination returned errors \n\n' + '\n'.join(
+ tracebacks))
+
if self.refit:
self.best_estimator_ = clone(base_estimator).set_params(**self.best_params_)
self.best_estimator_.fit(X, y, **fit_params)
--
GitLab