Abstract: Systems and methods for dynamically determining an optimal baseline algorithm for calculating lift values are disclosed. The system receives data associated with a control strategy, and then randomly selects a control location, a time period, and an item that may not be associated with the control strategy but meets the one or more criteria of the control strategy such as relevance and sales volume. Using the randomly selected inputs and a plurality of null baselines values determined by a plurality of null baseline algorithms, the system iteratively calculates a plurality of null lift values for each of the applied plurality of null baselines values to determine a likelihood for a false positive lift for each of the applied plurality of null baselines values. An optimal baseline algorithm is selected from the plurality of null baselines algorithms based on their corresponding likelihood of false positive lifts.

Abstract: Systems and methods for dynamically determining an optimal baseline algorithm for calculating lift values are disclosed. The system receives data associated with a control strategy, and then randomly selects a control location, a time period, and an item that may not be associated with the control strategy but meets the one or more criteria of the control strategy such as relevance and sales volume. Using the randomly selected inputs and a plurality of null baselines values determined by a plurality of null baseline algorithms, the system iteratively calculates a plurality of null lift values for each of the applied plurality of null baselines values to determine a likelihood for a false positive lift for each of the applied plurality of null baselines values. An optimal baseline algorithm is selected from the plurality of null baselines algorithms based on their corresponding likelihood of false positive lifts.

Abstract: Systems and methods for dynamically determining an optimal baseline algorithm for calculating lift values are disclosed. The system receives data associated with a control strategy, and then randomly selects a control location, a time period, and an item that may not be associated with the control strategy but meets the one or more criteria of the control strategy such as relevance and sales volume. Using the randomly selected inputs and a plurality of null baselines values determined by a plurality of null baseline algorithms, the system iteratively calculates a plurality of null lift values for each of the applied plurality of null baselines values to determine a likelihood for a false positive lift for each of the applied plurality of null baselines values. An optimal baseline algorithm is selected from the plurality of null baselines algorithms based on their corresponding likelihood of false positive lifts.

Abstract: The computer system and method described herein attempt to address the deficiencies by analyzing all relevant data points for each test and control location collectively determine outliers and then exclude the individual outlier data points from the data when analyzing an initiative during a relevant test period. Rather than exclude outliers at the site level, the particular time increment having the outlier data can be extracted and the site can remain in the analysis.

Abstract: The computer system and method described herein attempt to address the deficiencies by analyzing all relevant data points for each test and control location collectively determine outliers and then exclude the individual outlier data points from the data when analyzing an initiative during a relevant test period. Rather than exclude outliers at the site level, the particular time increment having the outlier data can be extracted and the site can remain in the analysis.

Abstract: Systems and methods for dynamically determining an optimal baseline algorithm for calculating lift values are disclosed. The system receives data associated with a control strategy, and then randomly selects a control location, a time period, and an item that may not be associated with the control strategy but meets the one or more criteria of the control strategy such as relevance and sales volume. Using the randomly selected inputs and a plurality of null baselines values determined by a plurality of null baseline algorithms, the system iteratively calculates a plurality of null lift values for each of the applied plurality of null baselines values to determine a likelihood for a false positive lift for each of the applied plurality of null baselines values. An optimal baseline algorithm is selected from the plurality of null baselines algorithms based on their corresponding likelihood of false positive lifts.