Note
Go to the end to download the full example code
Tuning Hyperparameters using Bayesian Search#
This example uses the fmri dataset, performs simple binary classification
using a Support Vector Machine classifier and analyzes the model.
References#
Waskom, M.L., Frank, M.C., Wagner, A.D. (2016). Adaptive engagement of cognitive control in context-dependent decision-making. Cerebral Cortex.
# Authors: Federico Raimondo <f.raimondo@fz-juelich.de>
# License: AGPL
import numpy as np
from seaborn import load_dataset
from julearn import run_cross_validation
from julearn.utils import configure_logging, logger
from julearn.pipeline import PipelineCreator
Set the logging level to info to see extra information.
configure_logging(level="INFO")
/home/runner/work/julearn/julearn/julearn/utils/logging.py:66: UserWarning: The '__version__' attribute is deprecated and will be removed in MarkupSafe 3.1. Use feature detection, or `importlib.metadata.version("markupsafe")`, instead.
vstring = str(getattr(module, "__version__", None))
2024-10-23 11:29:18,052 - julearn - INFO - ===== Lib Versions =====
2024-10-23 11:29:18,052 - julearn - INFO - numpy: 1.26.4
2024-10-23 11:29:18,052 - julearn - INFO - scipy: 1.14.1
2024-10-23 11:29:18,052 - julearn - INFO - sklearn: 1.5.2
2024-10-23 11:29:18,052 - julearn - INFO - pandas: 2.2.3
2024-10-23 11:29:18,052 - julearn - INFO - julearn: 0.3.5.dev16
2024-10-23 11:29:18,052 - julearn - INFO - ========================
Set the random seed to always have the same example.
np.random.seed(42)
Load the dataset.
df_fmri = load_dataset("fmri")
df_fmri.head()
Set the dataframe in the right format.
df_fmri = df_fmri.pivot(
index=["subject", "timepoint", "event"], columns="region", values="signal"
)
df_fmri = df_fmri.reset_index()
df_fmri.head()
Following the hyperparamter tuning example, we will now use a Bayesian search to find the best hyperparameters for the SVM model.
X = ["frontal", "parietal"]
y = "event"
creator1 = PipelineCreator(problem_type="classification")
creator1.add("zscore")
creator1.add(
"svm",
kernel=["linear"],
C=(1e-6, 1e3, "log-uniform"),
)
creator2 = PipelineCreator(problem_type="classification")
creator2.add("zscore")
creator2.add(
"svm",
kernel=["rbf"],
C=(1e-6, 1e3, "log-uniform"),
gamma=(1e-6, 1e1, "log-uniform"),
)
search_params = {
"kind": "bayes",
"cv": 2, # to speed up the example
"n_iter": 10, # 10 iterations of bayesian search to speed up example
}
scores, estimator = run_cross_validation(
X=X,
y=y,
data=df_fmri,
model=[creator1, creator2],
cv=2, # to speed up the example
search_params=search_params,
return_estimator="final",
)
print(scores["test_score"].mean())
2024-10-23 11:29:18,060 - julearn - INFO - Adding step zscore that applies to ColumnTypes<types={'continuous'}; pattern=(?:__:type:__continuous)>
2024-10-23 11:29:18,061 - julearn - INFO - Step added
2024-10-23 11:29:18,061 - julearn - INFO - Adding step svm that applies to ColumnTypes<types={'continuous'}; pattern=(?:__:type:__continuous)>
2024-10-23 11:29:18,061 - julearn - INFO - Setting hyperparameter kernel = linear
2024-10-23 11:29:18,061 - julearn - INFO - Tuning hyperparameter C = (1e-06, 1000.0, 'log-uniform')
2024-10-23 11:29:18,061 - julearn - INFO - Step added
2024-10-23 11:29:18,061 - julearn - INFO - Adding step zscore that applies to ColumnTypes<types={'continuous'}; pattern=(?:__:type:__continuous)>
2024-10-23 11:29:18,061 - julearn - INFO - Step added
2024-10-23 11:29:18,061 - julearn - INFO - Adding step svm that applies to ColumnTypes<types={'continuous'}; pattern=(?:__:type:__continuous)>
2024-10-23 11:29:18,061 - julearn - INFO - Setting hyperparameter kernel = rbf
2024-10-23 11:29:18,061 - julearn - INFO - Tuning hyperparameter C = (1e-06, 1000.0, 'log-uniform')
2024-10-23 11:29:18,061 - julearn - INFO - Tuning hyperparameter gamma = (1e-06, 10.0, 'log-uniform')
2024-10-23 11:29:18,061 - julearn - INFO - Step added
2024-10-23 11:29:18,061 - julearn - INFO - ==== Input Data ====
2024-10-23 11:29:18,061 - julearn - INFO - Using dataframe as input
2024-10-23 11:29:18,062 - julearn - INFO - Features: ['frontal', 'parietal']
2024-10-23 11:29:18,062 - julearn - INFO - Target: event
2024-10-23 11:29:18,062 - julearn - INFO - Expanded features: ['frontal', 'parietal']
2024-10-23 11:29:18,062 - julearn - INFO - X_types:{}
2024-10-23 11:29:18,062 - julearn - WARNING - The following columns are not defined in X_types: ['frontal', 'parietal']. They will be treated as continuous.
/home/runner/work/julearn/julearn/julearn/prepare.py:509: RuntimeWarning: The following columns are not defined in X_types: ['frontal', 'parietal']. They will be treated as continuous.
warn_with_log(
2024-10-23 11:29:18,062 - julearn - INFO - ====================
2024-10-23 11:29:18,062 - julearn - INFO -
2024-10-23 11:29:18,063 - julearn - INFO - = Model Parameters =
2024-10-23 11:29:18,063 - julearn - INFO - Tuning hyperparameters using bayes
2024-10-23 11:29:18,063 - julearn - INFO - Hyperparameters:
2024-10-23 11:29:18,063 - julearn - INFO - svm__C: (1e-06, 1000.0, 'log-uniform')
2024-10-23 11:29:18,063 - julearn - INFO - Hyperparameter svm__C is log-uniform float [1e-06, 1000.0]
2024-10-23 11:29:18,064 - julearn - INFO - Using inner CV scheme KFold(n_splits=2, random_state=None, shuffle=False)
2024-10-23 11:29:18,064 - julearn - INFO - Search Parameters:
2024-10-23 11:29:18,064 - julearn - INFO - cv: KFold(n_splits=2, random_state=None, shuffle=False)
2024-10-23 11:29:18,064 - julearn - INFO - n_iter: 10
2024-10-23 11:29:18,064 - julearn - INFO - ====================
2024-10-23 11:29:18,064 - julearn - INFO -
2024-10-23 11:29:18,065 - julearn - INFO - = Model Parameters =
2024-10-23 11:29:18,065 - julearn - INFO - Tuning hyperparameters using bayes
2024-10-23 11:29:18,065 - julearn - INFO - Hyperparameters:
2024-10-23 11:29:18,065 - julearn - INFO - svm__C: (1e-06, 1000.0, 'log-uniform')
2024-10-23 11:29:18,065 - julearn - INFO - svm__gamma: (1e-06, 10.0, 'log-uniform')
2024-10-23 11:29:18,065 - julearn - INFO - Hyperparameter svm__C is log-uniform float [1e-06, 1000.0]
2024-10-23 11:29:18,066 - julearn - INFO - Hyperparameter svm__gamma is log-uniform float [1e-06, 10.0]
2024-10-23 11:29:18,066 - julearn - INFO - Using inner CV scheme KFold(n_splits=2, random_state=None, shuffle=False)
2024-10-23 11:29:18,066 - julearn - INFO - Search Parameters:
2024-10-23 11:29:18,067 - julearn - INFO - cv: KFold(n_splits=2, random_state=None, shuffle=False)
2024-10-23 11:29:18,067 - julearn - INFO - n_iter: 10
2024-10-23 11:29:18,067 - julearn - INFO - ====================
2024-10-23 11:29:18,067 - julearn - INFO -
2024-10-23 11:29:18,067 - julearn - INFO - = Model Parameters =
2024-10-23 11:29:18,067 - julearn - INFO - Tuning hyperparameters using bayes
2024-10-23 11:29:18,067 - julearn - INFO - Hyperparameters list:
2024-10-23 11:29:18,067 - julearn - INFO - Set 0
2024-10-23 11:29:18,067 - julearn - INFO - svm__C: Real(low=1e-06, high=1000.0, prior='log-uniform', transform='identity')
2024-10-23 11:29:18,067 - julearn - INFO - set_column_types: [SetColumnTypes(X_types={})]
2024-10-23 11:29:18,067 - julearn - INFO - zscore: [StandardScaler()]
2024-10-23 11:29:18,068 - julearn - INFO - svm: [SVC(kernel='linear')]
2024-10-23 11:29:18,068 - julearn - INFO - Set 1
2024-10-23 11:29:18,068 - julearn - INFO - svm__C: Real(low=1e-06, high=1000.0, prior='log-uniform', transform='identity')
2024-10-23 11:29:18,068 - julearn - INFO - svm__gamma: Real(low=1e-06, high=10.0, prior='log-uniform', transform='identity')
2024-10-23 11:29:18,068 - julearn - INFO - set_column_types: [SetColumnTypes(X_types={})]
2024-10-23 11:29:18,068 - julearn - INFO - zscore: [StandardScaler()]
2024-10-23 11:29:18,068 - julearn - INFO - svm: [SVC()]
2024-10-23 11:29:18,068 - julearn - INFO - Hyperparameter svm__C as is Real(low=1e-06, high=1000.0, prior='log-uniform', transform='identity')
2024-10-23 11:29:18,068 - julearn - INFO - Hyperparameter set_column_types as is [SetColumnTypes(X_types={})]
2024-10-23 11:29:18,069 - julearn - INFO - Hyperparameter zscore as is [StandardScaler()]
2024-10-23 11:29:18,069 - julearn - INFO - Hyperparameter svm as is [SVC(kernel='linear')]
2024-10-23 11:29:18,069 - julearn - INFO - Hyperparameter svm__C as is Real(low=1e-06, high=1000.0, prior='log-uniform', transform='identity')
2024-10-23 11:29:18,069 - julearn - INFO - Hyperparameter svm__gamma as is Real(low=1e-06, high=10.0, prior='log-uniform', transform='identity')
2024-10-23 11:29:18,069 - julearn - INFO - Hyperparameter set_column_types as is [SetColumnTypes(X_types={})]
2024-10-23 11:29:18,069 - julearn - INFO - Hyperparameter zscore as is [StandardScaler()]
2024-10-23 11:29:18,069 - julearn - INFO - Hyperparameter svm as is [SVC()]
2024-10-23 11:29:18,069 - julearn - INFO - Using inner CV scheme KFold(n_splits=2, random_state=None, shuffle=False)
2024-10-23 11:29:18,070 - julearn - INFO - Search Parameters:
2024-10-23 11:29:18,070 - julearn - INFO - cv: KFold(n_splits=2, random_state=None, shuffle=False)
2024-10-23 11:29:18,070 - julearn - INFO - n_iter: 10
2024-10-23 11:29:18,078 - julearn - INFO - ====================
2024-10-23 11:29:18,078 - julearn - INFO -
2024-10-23 11:29:18,078 - julearn - INFO - = Data Information =
2024-10-23 11:29:18,078 - julearn - INFO - Problem type: classification
2024-10-23 11:29:18,079 - julearn - INFO - Number of samples: 532
2024-10-23 11:29:18,079 - julearn - INFO - Number of features: 2
2024-10-23 11:29:18,079 - julearn - INFO - ====================
2024-10-23 11:29:18,079 - julearn - INFO -
2024-10-23 11:29:18,079 - julearn - INFO - Number of classes: 2
2024-10-23 11:29:18,079 - julearn - INFO - Target type: object
2024-10-23 11:29:18,079 - julearn - INFO - Class distributions: event
cue 266
stim 266
Name: count, dtype: int64
2024-10-23 11:29:18,080 - julearn - INFO - Using outer CV scheme KFold(n_splits=2, random_state=None, shuffle=False) (incl. final model)
2024-10-23 11:29:18,080 - julearn - INFO - Binary classification problem detected.
0.656015037593985
It seems that we might have found a better model, but which one is it?
print(estimator.best_params_)
OrderedDict([('set_column_types', SetColumnTypes(X_types={})), ('svm', SVC()), ('svm__C', 0.0018082604408073564), ('svm__gamma', 1.6437581151471767), ('zscore', StandardScaler())])
Total running time of the script: (0 minutes 3.860 seconds)