Please enable JavaScript.
Coggle requires JavaScript to display documents.
PRODUCT RANKING (Intro (With a lot of data, using machine learning…
PRODUCT RANKING
Intro
With a lot of data, using machine learning algorithm is a solution
-
The algorithm should be simple, reliable, and explanable
At the moment, supervised learning algorithms have been tried. All of them satisfy the requirements of simplicity, reliability and explanability
-
Data
Source
-
Product characteristic
-
len of product title, etc...
product clicks, sales, impression
comparing to other product
shop characteristic
-
some statistics of shop like the number of sale, number of click
relative statistic of shop comparing to other shop in the same category, etc.
Intro
-
Come from different sources like tracking listing log, track listing click log, etc.