WIP: User preference prediction openday #53
Conversation
@Baschdl what do you mean? |
domain_scoring/domain_scoring.py
Outdated
| @@ -112,3 +114,44 @@ def _extract_data_labels(self, metapath_graph: MetaPathRatingGraph) -> (List[Tup | |||
| metapath_labels.append(LARGER) # > | |||
|
|
|||
| return metapath_pairs, metapath_labels | |||
|
|
|||
| def _test_score(self, x_test, y_test): | |||
| print('Test accuracy is {}'.format(self.classifier.score(X=self._preprocess(x_test), y=y_test))) | |||
There was a problem hiding this comment.
This doesn't always yield the accuracy as we say with the RandomForestRegressor
There was a problem hiding this comment.
You mean that this depends on the predictor?
There was a problem hiding this comment.
Yes, either we we don't call self.classifier.score and instead predict the labels with our classifier and evaluate it with the general scikit-learn accuracy() method or we print "Test score is..." but that's more or less useless if you don't know which score this is.
Like clean up things etc. |
Pull Request Test Coverage Report for Build 440
💛 - Coveralls |
…m/KDD-OpenSource/32de-python into userpreference-prediction-openday
|
broken build: We have to use a consistent filename for the datasets |
|
The (Same error occurs locally) |
|
|
|
Fixed in #59 |
# Conflicts: # requirements.txt
…m/KDD-OpenSource/32de-python into userpreference-prediction-openday * 'userpreference-prediction-openday' of https://github.com/KDD-OpenSource/32de-python: (40 commits) Set cryptography to working version Updated dataset paths in notebooks. Restructured rated datasets Add new structured rnn notebook. Add rnn notebook with high score. Add notebook for evaluating the ground truth from the KDD day Parametrize the algorithm used by the oracle Add prediction based on past mean of ratings to random sampler Remove plotting capabilities Fix use mean of squared error Add FlexibleOracle for arbitrary rating methods Fix use of only recently used ratings for GPR fitting Add simple experiments on gaussian process regressors Remove kernel optimizer Remove plotting from test for travis Add tests for exmperimental setups Refactor meta-path constructor Add tfidf vectorization Remove comments. Add dirty fix for CORS ...
@GittiHab What do we have to do to make this mergeable into our master?