Marketing Sales Prediction Algorithm using Prior Sales Data

Project Description

UPDATE 2-12-2015: At current we're in the process of reviewing bids that have already been submitted and not accepting new bids.

==============

Goals: We'd like to have an algorithm developed that can look at a large data set and identify our potential customers with a high degree of accuracy. The algorithm should also be able to identify non-potential customers with a high degree of accuracy.

This will allow us to lower our marketing costs and/or completely eliminate non-buyers from our marketing pool.

We have two data sets to be building an algorithm from and should be used to test the accuracy of the algorithm against.

Each data set of individuals is about 700,000-800,000 with about 4000 buyers in each.

There is a variety of parameters associated with each individual, from address to birthday, to high school type, to high school name and gender. Also beyond this, we're open to using data enhancement services at your recommendation (which we have available) to include household income, credit and/or any other parameters deemeded testable. The data enhancement service has about 1000 additional parameters which they can attach beyond what we have.

At the end of this, I'd like to have an algorithm I can feed new data into that will identify the most likely buyers based upon the two previous data sets such to tag a new live data set we'll be mailing and see if the algorithm performs in the wild.

Data will be provided in CSV format at project start time.