Auction Pricing {Laboratory}

Goal

Predict the sale price of a concert ticket on eBay to eventually buy hold and sale.

Motivated by

proposal as a hands-on experience for ML Class.

Features and their sources

• previous sales
  • eBay API e.g. with specific categories
  • average price ticket http://www.pollstar.com/atpDetail.aspx?SearchBy=A
• date of the concert
  • eventful API
• concert location
  • eventful API
• size of the room
  • http://upcoming.yahoo.com/services/api/event.search.php
• music style
  • ?
• estimated popularity
  • Twitter activity
  • Facebook fan page
  • http://www.ticketmaster.com via Facebook RSVP

Possibly also

• http://www.infochimps.com/search?view=list&price_category=&has_categories=&dataset_type=&order=score&tags=&query=concert

Example dataset (CSV)

price,date of sale,date of the concert,band name, concert location,size of the room,music style,estimated popularity
100,01/11/2011,06/11/2011,Texas,,10000,rock,
101,02/11/2011,06/11/2011,Texas,,10000,rock,
110,05/11/2011,06/11/2011,Texas,,10000,rock,
...

Tasks

1. gather and clean the dataset (scrapped or via API)
2. configure AI4R to compute the function parameters
3. provide a web interface for the result given features of the current page
4. browse the original website

See also

• Ruby with AI4R
• Octave
• AIClass#MachineLearning
• Seedea:Seedea/DATAmatrix
• remote_overlay.user.js to display the result
• ML and prediction markets
  • http://www.cs.ucla.edu/~jenn/pubs.html
  • http://homepages.inf.ed.ac.uk/amos/publications.html
  • http://www.stat.fsu.edu/~abarbu/