Seismic Pattern Recognition:  Predictive models can be used to detect seismic events, pick first breaks, filter spikes, etc.

Drilling: During drilling predictive models can be used to detect conditions in which the drill pipe might get stuck and determine the likelihood of un-sticking it.  Predictive models can also be used to monitor and control drilling mud properties.

Well Log Analysis:  Predictive models can integrate data from wireline logs with core samples to predict key reservoir properties, such as permeability, where core samples are unavailable.

Decline Curve Analysis:  It can be a time consuming and error-prone process to eye-ball the production curves of every well in an active field.  Predictive models can analyze production histories to forecast future decline rates and detect failing equipment.

Pump Card Diagnosis:  A dynamometer pump card is generated by monitoring the pump load string through its full cycle.  The shape of the card is useful for diagnosing various problems with the pump.  Predictive models can be used to automate pump card diagnosis in real-time.