我有以下具有(滞后)交互作用的ARDL模型
dynlm(ts_sales ~ trend(ts_sales) + L(ts_sales, 1:4) + season(ts_sales) +
L(ts_DIFF_main_variable, 0:1)*L(ts_V1, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_DIFF_V2, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_DIFF_V3, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_DIFF_V4, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_V5, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_DIFF_V6, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_V7, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_DIFF_V8, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_DIFF_V9, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_V10, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_DIFF_V11, 0:1) +
L(ts_DIFF_main_variable, 0:1)*L(ts_DIFF_V12, 0:1), start = c(2017, 2), end = c(2019, 52))
如何用多项式滞后分布限制滞后权重?可以使用dnlm包中的crossbasis()函数以某种方式做到这一点吗?