fsm_batch

fsm_batch(x, x_past, z_past, order, *, s_function='dopt', q_initial=None, eps=0.001, ties='random', intercept=True, standardize=True, pol_mat=None, w_pol=None, keep_criteria=False, rng=None)[source]

Allocate a new batch of units given units already allocated.

Each group starts from the information carried by its past members, and standardization and the ridge use past and new units together, so a sequence of batches behaves like one FSM run over a growing sample.

Parameters:

  • x (ndarray) – Covariates of the new units.

  • x_past (ndarray) – Covariates of the units already allocated.

  • z_past (ndarray) – Group labels of the units already allocated.

  • order (SelectionOrder | ndarray) – Selection order for the new units, one stage per new unit.

  • s_function (Literal['constant', 'dopt', 'aopt', 'negative_dopt', 'max_pc', 'min_pc', 'dopt_pc', 'max_average', 'min_average', 'dopt_average', 'marginal_var_sum']) – Selection function; see fsm().

  • q_initial (ndarray | None) – See fsm().

  • eps (float) – See fsm().

  • ties (Literal['random', 'smallest']) – See fsm().

  • intercept (bool) – See fsm().

  • standardize (bool) – See fsm().

  • pol_mat (ndarray | None) – See fsm().

  • w_pol (ndarray | None) – See fsm().

  • keep_criteria (bool) – See fsm().

  • rng (Generator | None) – See fsm().

Returns:

The allocation of the new units; selected indexes rows of x.

Raises:

ValueError – If the inputs do not line up.

Return type:

FSMResult