MIDASVAR._burn_for

MIDASVAR._burn_for(order)

Leading observations lost to the lags or to the weight window.

A window of midas_lags sub-periods ending in low-frequency period t reaches back ceil(midas_lags / period) - 1 periods, and the burn is the larger of that reach and the endogenous order.

Parameters:

order (int) – Endogenous order to compute the burn for.

Return type:

int

Returns:

The number of leading low-frequency rows the design cannot use.

Example

>>> import numpy as np
>>> x = np.arange(24.0).reshape(24, 1)
>>> model = MIDASVAR(np.zeros((8, 2)), x, order=1, period=3, midas_lags=7)
>>> model._burn_for(1), model._burn_for(3)
(2, 3)