Wednesday, February 29, 2012

Predict Opportunity Close Dates with Google Prediction API

User-added image
We ran the Opportunity Close Predictor with Google Prediction API challenge because, frankly, we think its super slick. The Google Prediction API a cloud-based machine learning tool that can help analyze data to add features to applications like recommendation systems, upsell opportunity analysis, spam detection, customer sentiment and much more.

Fortunately Google offers a number of REST client libraries for various languages so participants submitted code in Ruby, Grails and Google Go. It was interesting that we didn't have any submissions using Java or Python which run natively on Google App Engine.

We wanted to make the applications as data agnostic as possible so developers were not required to develop an app that was tightly integrated with, just to use some sort of Opportunity data. It was their choice to either integrated directly with or use an export of data. We suggested using data points like amount, close date, customer, products, sales rep, and stage to get them started in the data training process.

Besides that, the requirements for the challenge were pretty light. Just use the Google Prediction API to predict a close date and/or probability of closing for Opportunities in We received 4 great submissions and when you have time, you must watch some of these submission videos. Some really cool work by these guys.

peakpado - demo application & submission video
Ruby on Heroku
Place: 1
Money: $1000
wcheung - demo application & submission video
Grails on Cloud Foundry
Place: 2
Money: $1000
PaulKolbovich - demo application & submission video
Google Go on App Engine
Place: 3
Money: $500
Slava - demo application
Ruby on Heroku
Place: 4
Money: $250

No comments:

Post a Comment