Price prediction-based Automated Content generation
Price prediction-based Automated Content generation
Here’s one interesting use-case for automated content related to deals advice for buyers. Suppose you want to identify best deals for used car buyers. Typically, used car transaction sites have thousands of listings, and they provide multiple filters for a buyer to look at different options. On most sites, they don’t indicate which are better deals. Afterall, the site’s interest is to get as many cars sold as possible.
Cartoq, a leading content website in auto in India (annual uniques: 43mn), takes the used cars listings data from all websites, and then identifies the best deals among them. How? Using XGBoost, it has trained a model for predicting the “true price” of a used car, based on multiple features (age, km, brand etc). That true price is compared with the list price (sought by the seller). Cars with list price closer to true price then get selected over those with higher differences.
Every day, new car listings are parsed through the XGBoost-trained model, the better deals identified, and these get pushed into a content generation system that publishes listicles of best deals – based on city, car type, buyer budget etc. The automated system runs daily validation of deals that were selected, updated with fresh deals wherever the cars have been sold. So all the listicles get dynamically updated, as and when the listing data gets updated on the transaction sites.
Check out this sample content that was created, and gets updated periodically – powered by XGBoost-powered True Price model developed for Cartoq.
Image Source: Shutterstock