EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
1577. Stochastic Glance-View estimator
Invited abstract in session WA-57: Large-scale Speed-sensitive Network Optimization, stream Optimization at Amazon.
Wednesday, 8:30-10:00Room: S06 (building: 101)
Authors (first author is the speaker)
| 1. | Jean-Charles Croix
|
| Amazon | |
| 2. | Marc Bataillou Almagro
|
| Amazon | |
| 3. | Augustin Lombard
|
| Amazon | |
| 4. | Mehdi Bennaceur
|
| Amazon |
Abstract
The promised delivery speed enjoyed by Amazon's customers when glancing over the website's items (ASIN) is the result of a cascade of planning algorithms. At the core of this cascade is the modelling of the expected speed given a network design and inventory allocation. In this work, we present a scalable probabilistic approach to estimate the future glance-view speed, enabling us to query inventory aware speed variations in micro seconds. The close formulae derived in this work decouples inventory coverage from glance-view count forecasting, allowing us to answer valuable what-if scenarios regarding inventory placement, or new site launches. This in turn can guide Amazon's multi-million network design initiatives from short to long term horizons.
Keywords
- E-Commerce
- Machine Learning
Status: accepted
Back to the list of papers