Tutorial: F3Pipeline with Cache¶
This tutorial will guide you through a simple example of using F3Pipeline, including the Cache functionality.
1. Imports¶
First, we import the required classes and functions:
from framework3 import F3Pipeline
from framework3.plugins.filters import StandardScalerPlugin
from framework3.plugins.filters import ClassifierSVMPlugin
from framework3.plugins.metrics import F1
from framework3.base import XYData
from framework3.plugins.filters import Cached
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from rich import print
2. Data Preparation¶
We load the Iris dataset and split it into training and test sets:
# Load the Iris dataset
iris = load_iris()
X, y = iris.data, iris.target
# Split the data into training and test sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
# Crear objetos XYData
x_train = XYData('Iris dataset X train', 'dataset', X_train)
y_train = XYData('Iris dataset Y train', 'dataset', y_train)
x_test = XYData('Iris dataset X test', 'dataset', X_test)
y_test = XYData('Iris dataset Y test', 'dataset', y_test)
3. Creating the Pipeline with Cache¶
We create a pipeline that includes a cached StandardScalerPlugin and a ClassifierSVMPlugin:
pipeline = F3Pipeline(
filters=[
Cached(
filter=StandardScalerPlugin(),
cache_data=True,
cache_filter=True,
overwrite=False
),
ClassifierSVMPlugin()
],
metrics=[F1()]
)
print(pipeline)
F3Pipeline(
filters=[
Cached(filter=StandardScalerPlugin(), cache_data=True, cache_filter=True, overwrite=False, storage=None),
ClassifierSVMPlugin(C=1.0, kernel='linear', gamma='scale')
],
metrics=[F1(average='weighted')],
overwrite=False,
store=False,
log=False
)
4. Training the Model¶
We train the model with the training data:
____________________________________________________________________________________________________
Fitting pipeline...
****************************************************************************************************
* Cached({'filter': StandardScalerPlugin({}), 'cache_data': True, 'cache_filter': True, 'overwrite': False, 'storage': None}):
- El filtro StandardScalerPlugin({}) con hash 4f0150e0e11419085ce0f08ab077b7e5891f817b No existe, se va a entrenar.
- El filtro StandardScalerPlugin({}) Se cachea.
* Saving in local path: cache/StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b/model
* Saved !
- El dato XYData(_hash='f77f9d95466939988cdd6a13f0cb91260b94c99d', _path='StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b') No existe, se va a crear.
- El dato XYData(_hash='f77f9d95466939988cdd6a13f0cb91260b94c99d', _path='StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b') Se cachea.
* Saving in local path: cache/StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b/f77f9d95466939988cdd6a13f0cb91260b94c99d
* Saved !
* ClassifierSVMPlugin({'C': 1.0, 'kernel': 'linear', 'gamma': 'scale'}):
5. Prediction and Evaluation¶
We make predictions and evaluate the model:
# Make predictions
predictions = pipeline.predict(x_test)
# Evaluate the model
evaluation = pipeline.evaluate(x_test, y_test, predictions)
print("Evaluation results:", evaluation)
Predicting pipeline...
****************************************************************************************************
* Cached({'filter': StandardScalerPlugin({}), 'cache_data': True, 'cache_filter': True, 'overwrite': False, 'storage': None})
- El dato XYData(_hash='0403ab1f29eb3d1d59f857aa03f7af153d7ff357', _path='StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b') No existe, se va a crear.
- El dato XYData(_hash='0403ab1f29eb3d1d59f857aa03f7af153d7ff357', _path='StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b') Se cachea.
* Saving in local path: cache/StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b/0403ab1f29eb3d1d59f857aa03f7af153d7ff357
* Saved !
* ClassifierSVMPlugin({'C': 1.0, 'kernel': 'linear', 'gamma': 'scale'})
____________________________________________________________________________________________________
Evaluating pipeline...
____________________________________________________________________________________________________
Resultados de la evaluación:
{'F1': 0.9664109121909632}
6. Demonstrating Cache Usage¶
We run the pipeline again to demonstrate the use of cache:
print("Segunda ejecución (debería usar datos en caché):")
# Crear el pipeline con Cache
pipeline = F3Pipeline(
filters=[
Cached(
filter=StandardScalerPlugin(),
cache_data=True,
cache_filter=True,
overwrite=False,
),
KnnFilter(),
],
metrics=[F1()],
)
print(pipeline)
pipeline.fit(x_train, y_train)
predictions = pipeline.predict(x_test)
evaluation = pipeline.evaluate(x_test, y_test, predictions)
print("Resultados de la evaluación:", evaluation)
Segunda ejecución (debería usar datos en caché):
F3Pipeline(
filters=[
Cached(filter=StandardScalerPlugin(), cache_data=True, cache_filter=True, overwrite=False, storage=None),
KnnFilter(
n_neighbors=5,
weights='uniform',
algorithm='auto',
leaf_size=30,
p=2,
metric='minkowski',
metric_params=None,
n_jobs=None
)
],
metrics=[F1(average='weighted')],
overwrite=False,
store=False,
log=False
)
____________________________________________________________________________________________________
Fitting pipeline...
****************************************************************************************************
* Cached({'filter': StandardScalerPlugin({}), 'cache_data': True, 'cache_filter': True, 'overwrite': False,
'storage': None}):
- El filtro StandardScalerPlugin({}) Existe, se crea lambda.
- El dato XYData(_hash='f77f9d95466939988cdd6a13f0cb91260b94c99d', _path='StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b') Existe, se crea lambda.
* KnnFilter({'n_neighbors': 5, 'weights': 'uniform', 'algorithm': 'auto', 'leaf_size': 30, 'p': 2, 'metric': 'minkowski', 'metric_params': None, 'n_jobs': None}):
* Downloading: <_io.BufferedReader name='cache/StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b/f77f9d95466939988cdd6a13f0cb91260b94c99d'>
____________________________________________________________________________________________________
Predicting pipeline...
****************************************************************************************************
* Cached({'filter': StandardScalerPlugin({}), 'cache_data': True, 'cache_filter': True, 'overwrite': False, 'storage': None})
- El dato XYData(_hash='a2131ee7caeb76fd704b11b77d0456223b7e0437', _path='StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b') No existe, se va a crear.
- Existe un Lambda por lo que se recupera el filtro del storage.
* Downloading: <_io.BufferedReader name='cache/StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b/model'>
- El dato XYData(_hash='a2131ee7caeb76fd704b11b77d0456223b7e0437', _path='StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b') Se cachea.
* Saving in local path: cache/StandardScalerPlugin/4f0150e0e11419085ce0f08ab077b7e5891f817b/a2131ee7caeb76fd704b11b77d0456223b7e0437
* Saved !
* KnnFilter({'n_neighbors': 5, 'weights': 'uniform', 'algorithm': 'auto', 'leaf_size': 30, 'p': 2, 'metric': 'minkowski', 'metric_params': None, 'n_jobs': None})
____________________________________________________________________________________________________
Evaluating pipeline...
____________________________________________________________________________________________________
Resultados de la evaluación:
{'F1': 1.0}
In this second execution, you should notice that the StandardScalerPlugin uses the cached data, which may result in faster execution time.
Conclusion¶
This tutorial has demonstrated how to use F3Pipeline with the Cache functionality in Framework3. We've seen how to create a pipeline that includes a cached filter, how to train the model, make predictions, and evaluate performance. We've also shown how caching can improve performance in subsequent runs.