A way to generate predictions “offline” for multiple real-world data that needs inferencing and predictions. Batch mode predictions do not produce real-time results so this approach is best to use when decisions do not need to be made right away. The opposite of batch prediction is real-time prediction. Since batch mode systems can operate offline, they can utilize lower powered systems or systems not always connected to the network to provide inference.
