ordinalgbt.loss
===============

.. py:module:: ordinalgbt.loss


Functions
---------

.. autoapisummary::

   ordinalgbt.loss.dec_clip_y_pred
   ordinalgbt.loss.stack_zeros_ones
   ordinalgbt.loss.sigmoid
   ordinalgbt.loss.grad_sigmoid
   ordinalgbt.loss.hess_sigmoid
   ordinalgbt.loss.alpha2theta
   ordinalgbt.loss.theta2alpha
   ordinalgbt.loss.probas_from_y_pred
   ordinalgbt.loss.ordinal_logistic_nll
   ordinalgbt.loss.gradient_ordinal_logistic_nll
   ordinalgbt.loss.hessian_ordinal_logistic_nll
   ordinalgbt.loss.lgb_ordinal_loss


Module Contents
---------------

.. py:function:: dec_clip_y_pred(fun)

.. py:function:: stack_zeros_ones(a: numpy.ndarray, only_zeros=False) -> numpy.ndarray

   Stacks zeroes and ones on the left and rights part of the array
   Stacks horizontally zeros and ones on the left and right hand side of the array
   respectively. If only_zeros is true then it only stacks zeros. This is for the
   gradient which is zero at both ends of the sigmoid.
   e.g.::

       a = [[1,2],
           [3,4],
           [5,6]]
       returns
           [[0,1,2,1],
           [0,3,4,1],
           [0,5,6,1]]


   :param a: A 2D array to pad with zeroes and ones
   :type a: np.ndarray
   :param only_zeros: If true, then it pads with only zeroes
   :type only_zeros: bool, default False

   :rtype: np.ndarray


.. py:function:: sigmoid(z) -> numpy.ndarray

   Sigmoid
   Sigmoid implementation in numpy

   :param z:
   :type z: np.ndarray

   :returns: Sigmoid
   :rtype: np.ndarray


.. py:function:: grad_sigmoid(z) -> numpy.ndarray

   Gradient
   Gradient of the sigmoid

   :param z:
   :type z: np.ndarray

   :returns: Gradient of Sigmoid
   :rtype: np.ndarray


.. py:function:: hess_sigmoid(z) -> numpy.ndarray

   Hessian
   Hessian of the sigmoid
   Sigmoid implementation in numpy

   :param z:
   :type z: np.ndarray

   :returns: Hessian of sigmoid
   :rtype: np.ndarray


.. py:function:: alpha2theta(alpha)

.. py:function:: theta2alpha(theta)

.. py:function:: probas_from_y_pred(y_preds, theta)

   convers y_preds to probabilities


.. py:function:: ordinal_logistic_nll(y_true: numpy.ndarray, y_preds: numpy.ndarray, theta: numpy.ndarray)

   Ordinal Negative log lilelihood

   :param y_true: 1-D array with correct labels, starts from 0 and goes up to the number
                  of unique classes minus one (so unique values are 0,1,2 when dealing
                  with three classes)
   :type y_true: np.ndarray
   :param y_preds: 1-D array with predictions in latent space
   :type y_preds: np.ndarray
   :param theta: thresholds, 1-D array, size is the number of classes minus one.
   :type theta: np.ndarray

   :returns: logistic ordinal negative log likelihood
   :rtype: np.ndarray


.. py:function:: gradient_ordinal_logistic_nll(y_true: numpy.ndarray, y_preds: numpy.ndarray, theta: numpy.ndarray) -> numpy.ndarray

   Gradient of ordinal nll
   Gradient of the ordinal logistic regression with respect to the predictions

   :param y_true: 1-D array with correct labels, starts from 0 and goes up to the number
                  of unique classes minus one (so unique values are 0,1,2 when dealing
                  with three classes)
   :type y_true: np.ndarray
   :param y_preds: 1-D array with predictions in latent space
   :type y_preds: np.ndarray
   :param theta: thresholds, 1-D array, size is the number of classes minus one.
   :type theta: np.ndarray

   :returns: Gradient of logistic ordinal negative log likelihood
   :rtype: np.ndarray


.. py:function:: hessian_ordinal_logistic_nll(y_true: numpy.ndarray, y_preds: numpy.ndarray, theta: numpy.ndarray) -> numpy.ndarray

   Hessian of ordinal nll
   Hessian of the ordinal logistic regression with respect to the predictions

   :param y_true: 1-D array with correct labels, starts from 0 and goes up to the number
                  of unique classes minus one (so unique values are 0,1,2 when dealing
                  with three classes)
   :type y_true: np.ndarray
   :param y_preds: 1-D array with predictions in latent space
   :type y_preds: np.ndarray
   :param theta: thresholds, 1-D array, size is the number of classes minus one.
   :type theta: np.ndarray

   :returns: Hessian of logistic ordinal negative log likelihood
   :rtype: np.ndarray


.. py:function:: lgb_ordinal_loss(y_true: numpy.ndarray, y_pred: numpy.ndarray, theta: numpy.ndarray)

   Ordinal loss for lightgbm use
   The ordinal loss used in the lightgbm framework. Returns the
   gradient and hessian of the loss.

   :param y_true: 1-D array with correct labels, starts from 0 and goes up to the number
                  of unique classes minus one (so unique values are 0,1,2 when dealing
                  with three classes)
   :type y_true: np.ndarray
   :param y_preds: 1-D array with predictions in latent space
   :type y_preds: np.ndarray
   :param theta: thresholds, 1-D array, size is the number of classes minus one.
   :type theta: np.ndarray

   :returns: Gradient and Hessian of logistic ordinal negative log likelihood
   :rtype: (np.ndarray, np.ndarray)