From one of the white papers which explains deceptive marketing based on customer activity prediction Statisticians have always relied on calculus to understand and generate data on consumer statistics. Programs such as MATLAB have proved to be time consuming in building platforms for basic math, namely graphing, basic algebra, and calculus. We needed a better and easier way to calculate basic consumer statistics to learn about and understand consumer wants and opinions on price, and other aspects.
What it does
Predicts the customer behavior and help organizations take business decisions.
How we built it
We used archive banking data from University of California Irvine and created a data source in Amazon S3 bucket and connected it to AmazonML Model. AmazonML divides uses 70% of the data for training the model and evaluates the model with remaining data.
Challenges we ran into
Understanding quality metics like Area under curve. Finding the right dataset and use case.
Accomplishments that we're proud of
Just broke the biggest myth that ML is hard.
What we learned
Applying machine learning on financial data.
What's next for CustomerPrediction
Future enhancement includes Data Analytics driven by machine learning.