Content delivery based on corrective modeling techniques

A non-transitory computer-readable medium comprising computer program code, that, when executed by a processor, causes the processor to:

• receive, by an online system, a plurality of content items for delivery to users of the online system;
• extract a first set of features, each feature in the first set based on one or more of user profile attributes or content item attributes;
• apply the first set of features as input to a general model to generate a predicted output;
• extract a second set of features, each feature in the second set based on one or more of user profile attributes or content item attributes;
• apply the second set of features as input to a residual model to generate a predicted residual value;
• generate a corrected predicted output by combining the predicted output and the predicted residual value;
• select a content item from the plurality of content items for delivery to a user of the online system based on the corrected predicted output;
• and
• send the selected content item to the user of the online system.

A method comprising:

• receiving general training data comprising training examples, each training example comprising an output label that indicates whether an event has occurred;
• applying a first set of features as input to a general model to generate a predicted output that predicts whether the event has occurred;
• determining a residual by comparing the predicted output to an output label;
• training a residual model that generates a predicted residual, wherein the residual model is trained on residual training data that comprises the determined residual.