Negative log-likelihood for multinomial data under the alternative
multinom_log_lik.RdNegative log-likelihood for multinomial data under the alternative
Arguments
- beta_as_vector
A vector containing the values for all \(\beta_k\) for \(k = 1, \dots, J-1\) and \(\beta_{k0}, for k = 1, \dots, J\). In particular, this vector should be so that the \((J-1)(p+1)\) entries are \(\beta_{10}, \beta_{1}^{\top}, \beta_{20}, \beta_{2}^{\top}, \dots, \beta_{(J-1)0}, \beta_{J-1}^{\top}, \beta_{j0}\).
- Y
This should be the \(n x J\) data matrix of outcomes.
- X
This should be the \(n x p\) design matrix of covariates.