V0.1.6 - Identification of an electromechanical system¶
Example created by Wilson Rocha Lacerda Junior
More details about this data can be found in the following paper (in Portuguese): https://www.researchgate.net/publication/320418710_Identificacao_de_um_motorgerador_CC_por_meio_de_modelos_polinomiais_autorregressivos_e_redes_neurais_artificiais
pip install sysidentpy
import pandas as pd
import numpy as np
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sysidentpy.polynomial_basis import PolynomialNarmax
from sysidentpy.metrics import root_relative_squared_error
from sysidentpy.utils.generate_data import get_miso_data, get_siso_data
df1 = pd.read_csv('data/x_cc.csv')
df2 = pd.read_csv('data/y_cc.csv')
df2[5000:80000].plot(figsize=(10, 4))
<AxesSubplot:>
# we will decimate the data using d=500 in this example
x_train, x_valid = np.split(df1.iloc[::500].values, 2)
y_train, y_valid = np.split(df2.iloc[::500].values, 2)
Building a Polynomial NARX model¶
model = PolynomialNarmax(non_degree=2,
order_selection=True,
n_info_values=40,
extended_least_squares=False,
ylag=2, xlag=2,
info_criteria='bic',
estimator='recursive_least_squares'
)
model.fit(x_train, y_train)
yhat = model.predict(x_valid, y_valid)
rrse = root_relative_squared_error(y_valid, yhat)
print(rrse)
results = pd.DataFrame(model.results(err_precision=8,
dtype='dec'),
columns=['Regressors', 'Parameters', 'ERR'])
print(results)
ee, ex, extras, lam = model.residuals(x_valid, y_valid, yhat)
model.plot_result(y_valid, yhat, ee, ex, n=300)
C:\Users\wilso\miniconda3\envs\v0.1.4\lib\site-packages\sysidentpy\polynomial_basis\narmax.py:333: UserWarning: n_info_values is greater than the maximum number of all regressors space considering the chosen y_lag, u_lag, and non_degree. We set as 15
self.info_values = self.information_criterion(X, y)
0.08011571455094049
Regressors Parameters ERR
0 y(k-1) 1.3016 0.98600038
1 x1(k-1)^2 103.9300 0.00794805
2 y(k-2)^2 0.0000 0.00250906
3 x1(k-1)y(k-1) -0.1257 0.00143301
4 y(k-2) -0.5078 0.00102781
5 x1(k-1)y(k-2) 0.0560 0.00053520
6 x1(k-2) 349.8627 0.00027965
7 x1(k-2)y(k-1) -0.0840 0.00011221
8 x1(k-2)x1(k-1) -7.8186 0.00004547
9 x1(k-2)y(k-2) 0.0341 0.00003253
Testing different autoregressive models¶
from sklearn.neighbors import KNeighborsRegressor
from sklearn.svm import SVC, LinearSVC, NuSVC
from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import RandomForestRegressor, AdaBoostRegressor, GradientBoostingRegressor
from sklearn.naive_bayes import GaussianNB
from catboost import CatBoostRegressor
from sklearn.linear_model import BayesianRidge, ARDRegression
from sysidentpy.general_estimators import NARX
estimators = [
('KNeighborsRegressor', NARX(
base_estimator=KNeighborsRegressor(),
xlag=10,
ylag=10)),
('DecisionTreeRegressor', NARX(
base_estimator=DecisionTreeRegressor(),
xlag=10,
ylag=10)),
('RandomForestRegressor', NARX(
base_estimator=RandomForestRegressor(
n_estimators=200),
xlag=10,
ylag=10,
)),
('Catboost_NARX', NARX(
base_estimator=CatBoostRegressor(
iterations=800,
learning_rate=0.1,
depth=8),
xlag=10,
ylag=10,
non_degree=1,
fit_params={'verbose': False}
)),
('ARD_NARX', NARX(
base_estimator=ARDRegression(),
xlag=10,
ylag=10,
non_degree=2
)),
('Polynomial_NARX', PolynomialNarmax(non_degree=2,
order_selection=True,
n_info_values=50,
extended_least_squares=False,
ylag=10, xlag=10,
info_criteria='bic',
estimator='recursive_least_squares',
)),
]
resultados = {}
for nome_do_modelo, modelo in estimators:
resultados['%s' % (nome_do_modelo)] = []
modelo.fit(x_train, y_train)
yhat = modelo.predict(x_valid, y_valid)
result = root_relative_squared_error(y_valid, yhat)
resultados['%s' % (nome_do_modelo)].append(result)
print(nome_do_modelo, '%.3f' % np.mean(result))
03-08 22:53:49 - INFO - Training the model
03-08 22:53:49 - INFO - Creating the regressor matrix
03-08 22:53:49 - INFO - The regressor matrix have 21 features
03-08 22:53:49 - INFO - Done! Model is built!
KNeighborsRegressor 1.833
03-08 22:53:50 - INFO - Training the model
03-08 22:53:50 - INFO - Creating the regressor matrix
03-08 22:53:50 - INFO - The regressor matrix have 21 features
03-08 22:53:50 - INFO - Done! Model is built!
DecisionTreeRegressor 0.346
03-08 22:53:50 - INFO - Training the model
03-08 22:53:50 - INFO - Creating the regressor matrix
03-08 22:53:50 - INFO - The regressor matrix have 21 features
C:\Users\wilso\miniconda3\envs\v0.1.4\lib\site-packages\sysidentpy\general_estimators\narx.py:154: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples,), for example using ravel().
self.base_estimator.fit(X, y, **self.fit_params)
03-08 22:53:51 - INFO - Done! Model is built!
RandomForestRegressor 0.178
03-08 22:54:03 - INFO - Training the model
03-08 22:54:03 - INFO - Creating the regressor matrix
03-08 22:54:03 - INFO - The regressor matrix have 21 features
03-08 22:54:12 - INFO - Done! Model is built!
Catboost_NARX 0.179
03-08 22:54:13 - INFO - Training the model
03-08 22:54:13 - INFO - Creating the regressor matrix
03-08 22:54:13 - INFO - The regressor matrix have 231 features
C:\Users\wilso\miniconda3\envs\v0.1.4\lib\site-packages\sklearn\utils\validation.py:63: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().
return f(*args, **kwargs)
03-08 22:54:14 - INFO - Done! Model is built!
ARD_NARX 0.074
Polynomial_NARX 0.032
for aux_results, results in sorted(resultados.items(), key=lambda x: np.mean(x[1]), reverse=False):
print(aux_results, np.mean(results))
Polynomial_NARX 0.03208119098997185
ARD_NARX 0.07364659571266981
RandomForestRegressor 0.17752025859574144
Catboost_NARX 0.17923040407121107
DecisionTreeRegressor 0.3457206837080308
KNeighborsRegressor 1.833370478725381