3. Likelihood Maximization

Maximizes the likelihood function and thereby estimates the parameters of the model.

Output:

• Will create 3 files: one for the maximization process (_interations.csv), one for the final estimations (_final.csv), and a "parameter file" (_parameter_file.txt) that can directly be used as an input for a prediction run (this file is formatted like the input parameter file and does not contain the table contain the parameter settings)
• The interations file contains all likelihood evaluations of the likelihood maximization
• The final file contains the estimated error for the estimated parameters via a Hessian matrix. The Hessian is calculated using a range of finite differences that are set relative to the value of the respective parameter. i.e., epsilon=1e-2 corresponds to 1% of each parameter used for the Hessian matrix estimation. Finally, the number of data points, the total log-likelihood, the normalized log-likelihood, the used optimization algorithm, the set tolerance, and the search space (log/linear) are noted.
• The parameter file is in the format of the parameter file that was submitted to the code (see Sec. 2.1.2). It only contains the final estimation of each parameter. Thus, the parameters are treated as "fixed" when the code is run with this parameter file. This file can be used to run the prediction step directly (potentially on a different input file).
• As the maximization is run on the different segments independently, _segment followed by the segment index is added to the sample input file name in case there are multiple segments.