Training time

The machine-learning predictive model is produced in the training section of the machine-learning quest.

Building the predictive model consists of training, testing, and adjusting the model in an iterative process. This involves data preparation steps to clean and transform the raw data into a shape consumable by the machine-learning algorithm, processing the data through the algorithm to score and test the predictions and fine tune the model to achieve the best results. Post-processing steps can be applied to transform the results into a desired format.

Each training quest run is metered as training time, from the execution start to the execution finish.

The duration of the training time directly depends on the dataset size and the complexity of the model.

When the model is ready, it can be moved forward to the production section of the quest for deployment that will allow model consumption. Model consumption in production can be either batch production or an API endpoint for real-time inferences.