In [1]:

%load_ext autoreload
%autoreload 2

%load_ext autoreload %autoreload 2

Notebook: Scikit-learn Models¶

In [2]:

import matplotlib.pyplot as plt
import seaborn as sbn

sbn.set()

import matplotlib.pyplot as plt import seaborn as sbn sbn.set()

Load the dataset¶

In [3]:

from ceruleo.dataset.catalog.CMAPSS import CMAPSSDataset

from ceruleo.dataset.catalog.CMAPSS import CMAPSSDataset

2024-02-24 22:44:16.622494: I tensorflow/core/util/port.cc:113] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.
2024-02-24 22:44:16.624330: I external/local_tsl/tsl/cuda/cudart_stub.cc:31] Could not find cuda drivers on your machine, GPU will not be used.
2024-02-24 22:44:16.651060: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2024-02-24 22:44:16.651085: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2024-02-24 22:44:16.652011: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2024-02-24 22:44:16.656662: I external/local_tsl/tsl/cuda/cudart_stub.cc:31] Could not find cuda drivers on your machine, GPU will not be used.
2024-02-24 22:44:16.657264: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 AVX512F AVX512_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2024-02-24 22:44:17.314001: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
/home/luciano/venvs/ceruleo/lib/python3.11/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
  from .autonotebook import tqdm as notebook_tqdm

In [4]:

train_dataset = CMAPSSDataset(train=True, models='FD001')

test_dataset = CMAPSSDataset(train=False, models='FD001')[15:30]

train_dataset = CMAPSSDataset(train=True, models='FD001') test_dataset = CMAPSSDataset(train=False, models='FD001')[15:30]

Create a transformer for a dataset¶

In [5]:

from ceruleo.transformation.functional.transformers import Transformer
from ceruleo.transformation.features.selection import ByNameFeatureSelector
from ceruleo.transformation.functional.pipeline.pipeline import make_pipeline
from ceruleo.transformation.features.scalers import MinMaxScaler
from ceruleo.dataset.catalog.CMAPSS import sensor_indices
FEATURES = [train_dataset[0].columns[i] for i in sensor_indices]

from ceruleo.transformation.functional.transformers import Transformer from ceruleo.transformation.features.selection import ByNameFeatureSelector from ceruleo.transformation.functional.pipeline.pipeline import make_pipeline from ceruleo.transformation.features.scalers import MinMaxScaler from ceruleo.dataset.catalog.CMAPSS import sensor_indices FEATURES = [train_dataset[0].columns[i] for i in sensor_indices]

In [6]:

transformer = Transformer(
    pipelineX=make_pipeline(
        ByNameFeatureSelector(features=FEATURES), 
        MinMaxScaler(range=(-1, 1))

    ), 
    pipelineY=make_pipeline(
        ByNameFeatureSelector(features=['RUL']),  
    )
)

transformer = Transformer( pipelineX=make_pipeline( ByNameFeatureSelector(features=FEATURES), MinMaxScaler(range=(-1, 1)) ), pipelineY=make_pipeline( ByNameFeatureSelector(features=['RUL']), ) )

Split train-test-validation¶

In [7]:

from sklearn.model_selection import train_test_split

from sklearn.model_selection import train_test_split

In [8]:

train_dataset, val_dataset = train_test_split(train_dataset, train_size=0.9)

train_dataset, val_dataset = train_test_split(train_dataset, train_size=0.9)

In [9]:

len(train_dataset), len(val_dataset), len(test_dataset)

len(train_dataset), len(val_dataset), len(test_dataset)

Out[9]:

(90, 10, 15)

Models¶

Scikit-learn¶

In [10]:

import sklearn.pipeline as sk_pipeline
from sklearn.ensemble import RandomForestRegressor
from sklearn.compose import TransformedTargetRegressor
from ceruleo.models.sklearn import EstimatorWrapper, TimeSeriesWindowTransformer, CeruleoRegressor
from sklearn.linear_model import Ridge

import sklearn.pipeline as sk_pipeline from sklearn.ensemble import RandomForestRegressor from sklearn.compose import TransformedTargetRegressor from ceruleo.models.sklearn import EstimatorWrapper, TimeSeriesWindowTransformer, CeruleoRegressor from sklearn.linear_model import Ridge

In [11]:

regressor = CeruleoRegressor(
    TimeSeriesWindowTransformer(
        transformer,
        window_size=32,
        padding=True,
        step=1),   
    Ridge(alpha=15))

regressor.fit(train_dataset)

regressor = CeruleoRegressor( TimeSeriesWindowTransformer( transformer, window_size=32, padding=True, step=1), Ridge(alpha=15)) regressor.fit(train_dataset)

Out[11]:

CeruleoRegressor(regressor=Ridge(alpha=15),
                 ts_window_transformer=TimeSeriesWindowTransformer(transformer=<ceruleo.transformation.functional.transformers.Transformer object at 0x7d9eea07f2d0>,
                                                                   window_size=32))

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Val dataset results¶

In [12]:

fig, ax = plt.subplots(figsize=(17, 5))
ax.plot(regressor.predict(val_dataset))
ax.plot(regressor.ts_window_transformer.true_values(val_dataset))

fig, ax = plt.subplots(figsize=(17, 5)) ax.plot(regressor.predict(val_dataset)) ax.plot(regressor.ts_window_transformer.true_values(val_dataset))

Out[12]:

[<matplotlib.lines.Line2D at 0x7d9ee8207310>]

Test dataset results¶

In [13]:

fig, ax = plt.subplots(figsize=(17, 5))
ax.plot(regressor.predict(test_dataset))
ax.plot(regressor.ts_window_transformer.true_values(test_dataset))

fig, ax = plt.subplots(figsize=(17, 5)) ax.plot(regressor.predict(test_dataset)) ax.plot(regressor.ts_window_transformer.true_values(test_dataset))

Out[13]:

[<matplotlib.lines.Line2D at 0x7d9ee7f39810>]

Parameters grid search¶

In [14]:

from sklearn.model_selection import GridSearchCV
from sklearn.ensemble import RandomForestRegressor 
from ceruleo.models.sklearn import CeruleoMetricWrapper


transformer = Transformer(
    pipelineX=make_pipeline(
        ByNameFeatureSelector(features=FEATURES), 
        MinMaxScaler(range=(-1, 1))

    ), 
    pipelineY=make_pipeline(
        ByNameFeatureSelector(features=['RUL']),  
    )
)

regressor_gs = CeruleoRegressor(
    TimeSeriesWindowTransformer(
        transformer,
        window_size=32,
        padding=True,
        step=1),   
    Ridge(alpha=15)
)
grid_search = GridSearchCV(
    estimator=regressor_gs,
     param_grid={
        'ts_window_transformer__window_size': [5, 10],         
        'regressor': [Ridge(alpha=15), RandomForestRegressor(max_depth=5)]
    },
    scoring=CeruleoMetricWrapper('neg_mean_absolute_error'),
    cv=2
)


grid_search.fit(train_dataset)

from sklearn.model_selection import GridSearchCV from sklearn.ensemble import RandomForestRegressor from ceruleo.models.sklearn import CeruleoMetricWrapper transformer = Transformer( pipelineX=make_pipeline( ByNameFeatureSelector(features=FEATURES), MinMaxScaler(range=(-1, 1)) ), pipelineY=make_pipeline( ByNameFeatureSelector(features=['RUL']), ) ) regressor_gs = CeruleoRegressor( TimeSeriesWindowTransformer( transformer, window_size=32, padding=True, step=1), Ridge(alpha=15) ) grid_search = GridSearchCV( estimator=regressor_gs, param_grid={ 'ts_window_transformer__window_size': [5, 10], 'regressor': [Ridge(alpha=15), RandomForestRegressor(max_depth=5)] }, scoring=CeruleoMetricWrapper('neg_mean_absolute_error'), cv=2 ) grid_search.fit(train_dataset)

Out[14]:

GridSearchCV(cv=2,
             estimator=CeruleoRegressor(regressor=Ridge(alpha=15),
                                        ts_window_transformer=TimeSeriesWindowTransformer(transformer=<ceruleo.transformation.functional.transformers.Transformer object at 0x7d9ee81eac90>,
                                                                                          window_size=32)),
             param_grid={'regressor': [Ridge(alpha=15),
                                       RandomForestRegressor(max_depth=5)],
                         'ts_window_transformer__window_size': [5, 10]},
             scoring=<ceruleo.models.sklearn.CeruleoMetricWrapper object at 0x7d9eea3868d0>)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

In [15]:

fig, ax = plt.subplots(figsize=(17, 5))
ax.plot(regressor.ts_window_transformer.true_values(test_dataset), label='True values')
ax.plot(regressor.predict(test_dataset), label='Previous estimator')

ax.plot(grid_search.best_estimator_.predict(test_dataset), label='Best estimator')
ax.legend()

fig, ax = plt.subplots(figsize=(17, 5)) ax.plot(regressor.ts_window_transformer.true_values(test_dataset), label='True values') ax.plot(regressor.predict(test_dataset), label='Previous estimator') ax.plot(grid_search.best_estimator_.predict(test_dataset), label='Best estimator') ax.legend()

Out[15]:

<matplotlib.legend.Legend at 0x7d9ee7f3b190>