Automated system control with data analytics using doubly stochastic model

A method for controlling a robot to stock shelves in a warehouse, comprising:

• obtaining a time series of intermittent data from a sensor in the warehouse;
• evaluating the time series with a driver estimation process, wherein the driver estimation process estimates a driver for the time series with an iterative process that includes utilizing an Expectation/Maximization (EM) algorithm and a Markov Chain Monte Carlo (MCMC) algorithm to determine a probability model having set of model parameters within a model space;
• generating an estimated sequence of driver values based on the model parameters, wherein the estimated sequence of driver values forms the driver;
• processing the driver to generate future demand scenarios;
• and
• controlling the robot within the warehouse to stock shelves based on the future demand scenarios.

A system for re-ordering parts in an enterprise resource planning (ERP) system, comprising:

• a memory;
• and
• a processor coupled to the memory and configured to perform a process that includes:
• obtaining a time series of data associated with an intermittent demand for parts in the ERP system;
• evaluating the time series with a driver estimation process, wherein the driver estimation process estimates a driver for the time series with an iterative process that includes:
• providing a model space;
• utilizing an Expectation/Maximization (EM) algorithm and a Markov Chain Monte Carlo (MCMC) algorithm to determine a probability model having set of model parameters within the model space;
• and
• generating an estimated sequence of driver values based on the model parameters;
• processing the driver values to model future demand scenarios for the parts in the ERP system that are statistically similar but not identical to time series;
• and
• automatically re-ordering parts in the ERP system based on the future demand scenarios.