Machine Learning for Optimisation and Services : Example

Consider the service of managing in its entirety the print-copy-scan needs of an organisation, assuming one takes those needs as given. The service is realised through the installation and management of a set of devices on the set of sites of the customer. Many processes are involved in the provision of that service, offering many opportunities for optimisation, e.g.:

• choosing and positioning the devices on each site so as to minimise the overall number or cost of the device fleet but minimise at the same time the job execution delays and the inconvenience for users to get their job results;
• configuring each device so as to minimise its overall energy consumption through repeated activation of sleep periods but minimise at the same time the delays and energy consumption surges induced by device waking procedures;
• allocating the maintenance resources (call centres, customer service) so as to minimise personel costs but minimise at the same time the risk of falling below the thresholds defined in the service level agreement with the customer, such violations being generally associated with contractual financial penalties for the service provider;
• organising the supply chain which provides, for each device, the replacement for broken parts and exhausted consumables so as to minimise the useless storing of assets but minimise at the same time the down time of each device, and hence the unavailability for the consumer.