AppSignal — Track your app’s health, errors and performance with ease
Track your app’s health, errors and performance with ease
Promoted
Maker
Hey there PH, this is the first time I've be built anything like this, and I'd really appreciate some feedback.
Flashfed.com is built with Azure ML, Huginn and Wordpress. Huginn scrapes the sales, formats the data, connects to an Azure ML web service, and feeds the data to a wordpress install using the pluginhunt theme. So far, this is my MVP.
Machine learning was actually an after thought—originally I just wanted to build an aggregator, but using ML seemed like a good way to overcome the 'chicken or the egg' problem all community oriented services endure.
For those curious about how Flashfed uses machine learning, I trained a least squares linear regression model to return a score between 1-50 (most fall between 20-30) based on previous sales' social sharing metrics, using n-grams (unigrams) feature hashing in Azure ml. Basically, it scores the sale based on the words used in it's title.
