Watchalytics

• 

  Dashboard example

• 

  Weekly view of genre distribution

• 

  Hourly view of a single day's viewing

What it does

Uses a custom time-based decay algorithm to track TV watching habits based on a number of factors. The data is analysed down to the hourly level to be able to provide insight and suggest new programmes to watch within small windows. Bored on a Sunday afternoon? This can probably give you your Sunday fix.

How we built it

Pre-processed and cleaned a lot of the data before running the algorithm over the bulk of the data set to train and refine it, then some further testing. Built a Node.JS app on top to query and serve the collected data from MongoDB.

Challenges we ran into

There's a lot of noise in the original data, and a lot of the fields were quite unreliable.

Accomplishments that we're proud of

The algorithm seems to work, needs more input data to really make it shine.

What we learned

Processing millions of rows of data on a tiny VM isn't fast. (This wasn't a surprise).

What's next for Watchalytics

More data please

Built With

• beer
• coffee
• mongodb
• mountain-dew
• node.js
• php
• pp
• sky-viewing-data
• tea