System for control group optimization to identify optimal baseline algorithm

A computer-implemented method comprising:

• receiving, by an analytics server executing an artificial intelligence computer model having a plurality of neural network nodes, an input from a user interface comprising data associated with a control location, wherein the data associated with the control location comprises data associated with a control strategy comprising a pre-period, a post-period, and an item associated with the control strategy, wherein the post-period corresponds to a time-period during a first promotion period;
• randomly selecting, by the analytics server, a null test location, wherein the null test location is selected from a group of one or more locations not including the control location;
• randomly selecting, by the analytics server, a null time period, wherein the null time period is not based on the pre-period or the post-period;
• generating and transmitting, by the analytics server, a first instruction to a database to receive data associated with the item;
• upon transmitting the first instruction, receiving, by the analytics server and from the database, data associated with a plurality of items of a first item category of the item, the data associated with the plurality of items of the first item category of the item comprising sales data from the first promotion period;
• identifying, by the analytics server, a second item category responsive to determining a plurality of items of the second item category satisfies a criteria for relevance and sales volume compared to the plurality of items of the first item category, the analytics server determining the plurality of items of the second item category satisfies the criteria for sales volume by comparing the sales data from the first promotion period with sales data of the plurality of items of the second item category of a second post-period corresponding to a second time period during a second promotion period;
• randomly selecting, by the analytics server, a null item from the plurality of items of the second item category, the plurality of items of the second item category not including the item;
• iteratively executing, by the analytics server, each neural network node to calculate a plurality of null lift values using the randomly selected null time period, the randomly selected null test location, the randomly selected null item, and a plurality of null baselines determined based on a plurality of null baseline algorithms;
• wherein with each iteration the analytics server trains the artificial intelligence computer model by adjusting a weight of one or more neural network nodes associated with at least one of the randomly selected null time period, the randomly selected null test location, or the randomly selected null item;
• identifying, by the analytics server, a first subset of null baseline algorithms where each null baseline algorithm within the first subset of null baseline algorithms does not have a false positive lift based on the plurality of null lift values corresponding to the plurality of null baseline algorithms;
• selecting, by the analytics server, an optimal null baseline algorithm from the first subset of null baseline algorithms based on a rank of each null baseline algorithm within the first subset of null baseline algorithms, the rank of each null baseline algorithm based upon a corresponding null lift value, wherein the rank associated with each of the first subset of null baselines algorithms is adjusted upward when the corresponding null lift value is closer to 0;
• executing, by the analytics server the selected null baseline algorithm configured to calculate an optimal control strategy lift value based on the pre-period, the post-period, and the item associated with the control strategy;
• and
• generating and executing, by the analytics server, a second instruction causing display of the optimal control strategy lift value on the user interface.