flexcode.post_processing

Module Contents

Functions

normalize(cde_estimates[, tol, max_iter])

Normalizes conditional density estimates to be non-negative and

_normalize(density[, tol, max_iter])

Normalizes a density estimate to be non-negative and integrate to

sharpen(cde_estimates, alpha)

Sharpens conditional density estimates.

choose_sharpen(cde_estimates, z_grid, true_z, alpha_grid)

Chooses the sharpen parameter by minimizing cde loss.

remove_bumps(cde_estimates, delta)

Removes bumps in conditional density estimates

_remove_bumps(density, delta)

Removes bumps in conditional density estimates.

choose_bump_threshold(cde_estimates, z_grid, true_z, ...)

Chooses the bump threshold which minimizes cde loss.
normalize(cde_estimates, tol=1e-06, max_iter=200)[source]

Normalizes conditional density estimates to be non-negative and integrate to one.

Assumes densities are evaluated on the unit grid.

Parameters:

  • cde_estimates – a numpy array or matrix of conditional density estimates.

  • tol – float, the tolerance to accept for abs(area - 1).

  • max_iter – int, the maximal number of search iterations.

Returns:

the normalized conditional density estimates.

Return type:

numpy array or matrix.

_normalize(density, tol=1e-06, max_iter=500)[source]

Normalizes a density estimate to be non-negative and integrate to one.

Assumes density is evaluated on the unit grid.

Parameters:

  • density – a numpy array of density estimates.

  • z_grid – an array, the grid points at the density is estimated.

  • tol – float, the tolerance to accept for abs(area - 1).

  • max_iter – int, the maximal number of search iterations.

Returns:

the normalized density estimate.

Return type:

numpy array.

sharpen(cde_estimates, alpha)[source]

Sharpens conditional density estimates.

Assumes densities are evaluated on the unit grid.

Parameters:

  • cde_estimates – a numpy array or matrix of conditional density estimates.

  • alpha – float, the exponent to which the estimate is raised.

Returns:

the sharpened conditional density estimate.

Return type:

numpy array or matrix.

choose_sharpen(cde_estimates, z_grid, true_z, alpha_grid)[source]

Chooses the sharpen parameter by minimizing cde loss.

Parameters:

  • cde_estimates – a numpy matrix of conditional density estimates

  • true_z – an array of the true z values corresponding to the cde_estimates.

  • alpha_grid – an array of candidate sharpen parameter values.

Returns:

the sharpen parameter value from alpha_grid which minimizes cde loss.

Return type:

float

remove_bumps(cde_estimates, delta)[source]

Removes bumps in conditional density estimates

Assumes that cde_estimates are on the unit grid.

Parameters:

  • cde_estimates – a numpy array or matrix of conditional density estimates.

  • delta – float, the threshold for bump removal

Returns:

the conditional density estimates with bumps removed

Return type:

numpy array or matrix

_remove_bumps(density, delta)[source]

Removes bumps in conditional density estimates.

Assumes estimates are on the unit grid.

Parameters:

  • density – a numpy array of conditional density estimate.

  • delta – float, the threshold for bump removal.

Returns:

the conditional density estimate with bumps removed.

Return type:

numpy array.

choose_bump_threshold(cde_estimates, z_grid, true_z, delta_grid)[source]

Chooses the bump threshold which minimizes cde loss.

Parameters:

  • cde_estimates – a numpy array or matrix of conditional density estimates.

  • z_grid – an array, the grid points at which the density is estimated.b

  • true_z – the true z values corresponding to the conditional

denstity estimates. :param delta_grid: an array of candidate bump threshold values :returns: the bump threshold value from delta_grid which minimizes CDE loss :rtype: float