7.2 Data Preparation

All estimation quantlets in the XploRe quantlib gam expect at least two input parameters:

t
$ n \times p$ matrix, continuous explanatory variables for nonparametric part
y
$ n\times q$ vector (more exactly a matrix of $ q$ vectors), observed responses

For partially linear models additionally:

x
$ n\times d$ matrix, explanatory variables for linear part (at least the discrete variables).

There should be no vector of ones concatenated to the data. A constant is contained automatically in the nonparametric estimate for $ m \left( t \right)$.

Neither the matrix x nor the vector y should contain missing values (NaN) or infinite values (Inf, -Inf). Those should be identified by 13996 isNumber and removed by 13999 paf (or replaced by something reasonable) before using GAM quantlets.