search

gasparrini / dlnm

R package dlnm
68 stars 13 forks source link

Crossbasis for an incomplete series #10

Closed YZYEDDY closed 2 years ago

YZYEDDY commented 2 years ago

Hi Antonio,

In the crossbasis, x must be a complete series. What if some observations are missing? Are we still able to establish a DLNM?

gasparrini commented 2 years ago

Dear YZYEDDI,

as explained on the help page of crossbasis(), the first argument 'x' should be a time series of equally-spaced and complete observations, or alternatively a matrix of exposure histories with rows representing complete lagged exposure for each observation. Missing observations must be included (as missing), asthe series must be complete. Bear in mind that for any missing observation many missings can be produced in the transformed variables, due to the lagging.

I hope this helps.

best -Antonio Gasparrini