Transform

Utility function to transform features and target

Target variable

---

source

log_transform_y

 log_transform_y (data:tuple)

Log-10 transform of the target value

Takes and returns all (X, y, X_id, depth_order) tuple to be able to pipe the function

	Type	Details
data	tuple	(X, y, X_id, depth_order)

Features (spectra)

---

source

TakeDerivative

 TakeDerivative (window_length=11, polyorder=1, deriv=1)

Creates scikit-learn derivation custom transformer

Args: window_length: int, optional Specify savgol filter smoothing window length

polyorder: int, optional
    Specify order of the polynom used to interpolate derived signal

deriv: int, optional
    Specify derivation degree

Returns: scikit-learn custom transformer

src_dir = 'test'
fnames = ['spectra-features-smp.npy', 'spectra-wavenumbers-smp.npy', 
          'depth-order-smp.npy', 'target-smp.npy', 
          'tax-order-lu-smp.pkl', 'spectra-id-smp.npy']

X, X_names, depth_order, y, tax_lookup, X_id = load_kssl(src_dir, fnames=fnames)

tfm = TakeDerivative()
plot_spectra(tfm.fit_transform(X), X_names, figsize=(12,4))

---

source

SNV

 SNV ()

Creates scikit-learn SNV custom transformer

Args: None

Returns: scikit-learn custom transformer

tfm = SNV()
plot_spectra(tfm.fit_transform(X), X_names, figsize=(12,4))

class Center(BaseEstimator, TransformerMixin):
    """Creates scikit-learn Centering custom transformer

    Args:
        None

    Returns:
        scikit-learn custom transformer
    """
    def __init__(self):
        pass

    def fit(self, X, y=None):
        return self

    def transform(self, X, y=None):
        return X - np.mean(X, axis=1).reshape(-1, 1)

tfm = Center()
plot_spectra(tfm.fit_transform(X), X_names, figsize=(12,4))

---

source

DropSpectralRegions

 DropSpectralRegions (wavenumbers, regions=[2389, 2269])

Creates scikit-learn custom transformer dropping specific spectral region(s)

Args: wavenumbers: list List of wavenumbers where absorbance measured

regions: list
    List of region(s) to drop

Returns: scikit-learn custom transformer

tfm = DropSpectralRegions(X_names, regions=CO2_REGION)
plot_spectra(tfm.fit_transform(X), X_names, figsize=(12,4))