Installation¶

Install the current PyPI release:
```
$ pip install pylwl==0.2.0
```

Install directly from source code:

$ git clone https://github.com/thieu1995/PyLWL.git
$ cd PyLWL
$ python setup.py install

In case, you want to install the development version from Github:
```
$ pip install git+https://github.com/thieu1995/PyLWL
```

After installation, you can import it as any other Python module:

$ python
>>> import pylwl
>>> pylwl.__version__

Examples¶

In this section, we will explore the usage of the Adam-based Gradient Optimizer for Locally Weighted Regression:

from sklearn.datasets import load_diabetes
from pylwl import Data, GdLwRegressor


## Load data object
X, y = load_diabetes(return_X_y=True)
data = Data(X, y)

## Split train and test
data.split_train_test(test_size=0.2, random_state=2, inplace=True)
print(data.X_train.shape, data.X_test.shape)

## Scaling dataset
data.X_train, scaler_X = data.scale(data.X_train, scaling_methods=("standard", "minmax"))
data.X_test = scaler_X.transform(data.X_test)

data.y_train, scaler_y = data.scale(data.y_train, scaling_methods=("minmax"))
data.y_train = data.y_train.ravel()
data.y_test = scaler_y.transform(data.y_test.reshape(-1, 1))

## Train and test
model = GdLwRegressor(kernel="gaussian", tau=1.0, epochs=20,
                      optim="Adam", optim_paras=None, seed=42, verbose=True, device=None)
model.fit(data.X_train, data.y_train)

## Predict
print("Predicted:", model.predict(data.X_test))

## Calculate some metrics
print(model.score(X=data.X_test, y=data.y_test))
print(model.scores(X=data.X_test, y=data.y_test, list_metrics=["MAE", "RMSE", "R", "NNSE", "KGE", "R2"]))
print(model.evaluate(y_true=data.y_test, y_pred=model.predict(data.X_test), list_metrics=["MAE", "RMSE", "R", "NNSE", "KGE", "R2"]))

A real-world dataset contains features that vary in magnitudes, units, and range. We would suggest performing normalization when the scale of a feature is irrelevant or misleading. Feature Scaling basically helps to normalize the data within a particular range.