Predictive push logistic using runtime monitoring of hidden and visible data

Abstract

This report described a general purpose predictive logistics software package based on three primary components: 1. A spreadsheet of ship related orders. 2. A Probabilistic Temporal Finite State-Machine (PTFSM) automatically learned from that data. 3. (Optional) Expert rules written in English. The deliverable tool predicts orders per Ship-ItemType pairs. It has two main prediction modes: a. Predict a probability of an order (for one or more Ship-ItemType pairs) to be required in n weeks. b. Predict the number of weeks required for the probability of an order (for one or more Ship-ItemType pairs) to exceed some given value p.