import evaluations dataloader = MsdBbLoader( hits_file_path='/storage/nas3/datasets/music/billboard/msd_bb_matches.csv', non_hits_file_path= '/storage/nas3/datasets/music/billboard/msd_bb_non_matches.csv', features_path='/storage/nas3/datasets/music/billboard', non_hits_per_hit=1, features=[ *common.hl_list(), ], label='peak', nan_value=150, random_state=42, ) pipeline = Pipeline([ ('scale', MinMaxScaler()), ('linreg', Lasso(alpha=1.0, normalize=False)), ]) evaluator = GridEvaluator( parameters={}, grid_parameters=evaluations.grid_parameters(), ) result_handlers = [ result_handlers.print_gridsearch_results, ]
from sklearn.pipeline import Pipeline from sklearn.preprocessing import StandardScaler from nlp4musa2020.dataloaders import vectorizer from nlp4musa2020.dataloaders.alf200k import ALF200KLoader from nlp4musa2020.dataloaders.alf200k import genre_target_labels import nlp4musa2020.evaluators as evaluators dataloader = ALF200KLoader( path='data/processed/dataset-lfm-genres.pickle', load_feature_groups=[], text_vectorizers=vectorizer.lda(), target=genre_target_labels(), ) pipeline = Pipeline([ ('scaler', StandardScaler()), ('model', RandomForestClassifier(n_jobs=-1)), ]) evaluator = GridEvaluator( parameters={ 'model__n_estimators': [10, 100, 300], }, grid_parameters=evaluators.grid_parameters_genres(), ) result_handlers = [ result_handlers.print_gridsearch_results, ]
from nlp4musa2020.dataloaders.vectorizer import tfidf import nlp4musa2020.evaluators as evaluators from nlp4musa2020.models.simplenn_genre import SimpleGenreNN dataloader = ALF200KLoader('data/processed/dataset-lfm-genres.pickle', load_feature_groups=[], text_vectorizers=tfidf(), target=genre_target_labels()) pipeline = Pipeline([ ('scaler', StandardScaler()), ('model', SimpleGenreNN(epochs=50)), ]) evaluator = GridEvaluator( parameters={ 'model__dense_sizes': [ (32, 32), (64, 64), ], 'model__dropout_rate': [ 0.1, ], }, grid_parameters=evaluators.grid_parameters_genres(), ) result_handlers = [ result_handlers.print_gridsearch_results, ]
hits_file_path='/storage/nas3/datasets/music/billboard/msd_bb_matches.csv', non_hits_file_path= '/storage/nas3/datasets/music/billboard/msd_bb_non_matches.csv', features_path='/storage/nas3/datasets/music/billboard', non_hits_per_hit=1, features=[ *common.all_list(), ], label='peak', nan_value=150, random_state=42, ) pipeline = Pipeline([ ('scale', MinMaxScaler()), ('wide_and_deep', WideAndDeep(features=dataloader.feature_indices)), ]) evaluator = GridEvaluator( parameters={ 'wide_and_deep__epochs': [10, 50, 100, 200], 'wide_and_deep__batch_normalization': [False, True], 'wide_and_deep__dropout_rate': [None, 0.25, 0.5], }, grid_parameters=evaluations.grid_parameters(), ) result_handlers = [ result_handlers.print_gridsearch_results, ]
dataloader = ALF200KLoader(path='data/processed/dataset-lfm-genres.pickle', load_feature_groups=[ 'explicitness', ], text_vectorizers=None, target=[ 'alternative', 'blues', 'country', 'dance', 'electronic', 'funk', 'hip hop', 'indie', 'jazz', 'metal', 'pop', 'punk', 'rap', 'rnb', 'rock', 'soul' ]) pipeline = Pipeline([ ('scaler', StandardScaler()), ('model', KNeighborsClassifier(n_jobs=-1, algorithm='ball_tree')), ]) evaluator = GridEvaluator( parameters={ 'model__n_neighbors': [3, 4, 5, 10], 'model__weights': ['distance'], 'model__p': [1, 2], }, grid_parameters=evaluators.grid_parameters_genres(), ) result_handlers = [ result_handlers.print_gridsearch_results, ]
dataloader = ALF200KLoader('data/processed/dataset-lfm-genres.pickle', load_feature_groups=[ 'rhymes', ], text_vectorizers=None, target=genre_target_labels()) pipeline = Pipeline([ ('scaler', StandardScaler()), ('model', MultiOutputClassifier(LinearSVC())), ]) evaluator = GridEvaluator( parameters={ 'model__estimator__C': [ 0.1, 0.5, 1.0, 2.0, 5.0, ], 'model__estimator__loss': ['epsilon_insensitive'], }, grid_parameters=evaluators.grid_parameters_genres(), ) result_handlers = [ result_handlers.print_gridsearch_results, ]
dataloader = MsdBbLoader( hits_file_path='/storage/nas3/datasets/music/billboard/msd_bb_matches.csv', non_hits_file_path= '/storage/nas3/datasets/music/billboard/msd_bb_non_matches.csv', features_path='/storage/nas3/datasets/music/billboard', non_hits_per_hit=1, features=[ *common.hl_list(), *common.ll_list(), ], label='peak', nan_value=150, random_state=42, ) pipeline = Pipeline([ ('scale', MinMaxScaler()), ('logreg', LogisticRegression(multi_class='auto', solver='lbfgs')), ]) evaluator = GridEvaluator( parameters={ 'logreg__C': [1.0], }, grid_parameters=evaluations.grid_parameters(), ) result_handlers = [ result_handlers.print_gridsearch_results, ]
from nlp4musa2020.dataloaders.alf200k import ALF200KLoader, genre_target_labels import nlp4musa2020.evaluators as evaluators from nlp4musa2020.models.simplenn_genre import SimpleGenreNN dataloader = ALF200KLoader('data/processed/dataset-lfm-genres.pickle', load_feature_groups=[ 'audio', ], text_vectorizers=None, target=genre_target_labels()) pipeline = Pipeline([ ('scaler', StandardScaler()), ('model', DummyClassifier(strategy="constant", constant=[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0])), #rock ]) evaluator = GridEvaluator( parameters={ "model__random_state": [42], }, grid_parameters=evaluators.grid_parameters_genres(), ) result_handlers = [ result_handlers.print_gridsearch_results, ]