“Dr. Etzioni, a computer scientist at the University of Washington in Seattle who has founded four firms in all, says Decide relies on three main data sources: pricing data, news and rumours, and technical specifications. Pricing data comes from a variety of sources. Most are the company’s trade secret, though they always include current prices of goods and sales data. The model also uses feedback about how its predictions fare over time to fine-tune their probability estimates. With news and chatter, Decide scores sites by how accurate their scoops are for particular categories of goods. The algorithm discounts rumour-mongers and gives a greater weight to reliable sources. So far, the firm has amassed a year’s worth of data, many thousands of gigabytes in total.
“These reveal unexpected consumer behaviour. For example …”
To read the exciting conclusion, check out the full article in The Economist here.