Scientific system and method for optimizing television advertising

A system, for training a neural network for optimizing placement of select advertising in streaming content, comprising:

• a processor;
• a memory storing executable code, wherein the executable code drives the processor to:
• collect different data sets for a plurality of different attributes to optimize media advertising and drive multiple performance outcomes, wherein the different data sets are for at least a size, a frequency of delivery in a given period, a channel or connected-television application, a placement, a sequence, a particular time-of-day of display of advertising, one or more previous website visitors, a third-party data segment, and an element within a tv creative;
• identify at least one outcome, from the multiple performance outcomes, in the streaming content displayed on an external device;
• train a neural network in an initial phase with the different data sets;
• apply a feedback engine that provides feedback data collected on the different data sets and stored in a database;
• train the neural network in a subsequent phase, based on the feedback data on the different data sets stored in the database;
• obtain the select advertising from one or more advertising sources based on outcomes defined by the neural network and delivering the select advertising to a connected-television and over-the-top delivery mechanism for displaying the select advertising within the streaming content;
• and
• generate a household effectiveness graph, wherein the household effectiveness graph provides additional data for optimizing delivery of the select advertising, including at least one of postal data, latitude or longitude data, IP type, age, gender, and household income.