Manual: 4.4. Train the Model

Training the model is an automatic process. You can trigger the training of the model either via the APO Wizard or via the APO Menu by selecting the Model option.

The form will ask for you to select the method you would like to be applied. Both choices will delete the current model and create a new one from the historical data. Method Model only will just do that and nothing more. It will therefore keep all the suggestions already made and all the manually entered reactions from the operators. This option is the right one for you if you have had APO in active use for some time and you merely want to update the model but retain the suggestions made in the past. Method full recomputation will also delete all suggestions made along with any manual reactions to suggestions. It will then recreate them all after the model is finished. This is the right option to choose if you are deploying APO, using it for the first time or in the process of fine-tuning the model.

The time limit can be used to speed up the recomputation of all historical suggestions by only computing them from the begining of known history to the specified point in time. Generally, we do not recommend using this option and recommend modeling the entire dataset.

Please note that training can take a substantial period of time. The amount of time is directly and linearly proportional to the number of data points in the database. The process is dominated by the read/write speed of database operations and thus the speed of the hard drive. Please be patient.

At the end of training, the model is saved in the database. If you have selected full recomputation, you will be able to look at the newly computed suggestions and assess model quality. We strongly recommend that you do so.