Automatic Historical Feature Generation through Tree-based Method in Ads Prediction

12/31/2020
by   Hongjian Wang, et al.
0

Historical features are important in ads click-through rate (CTR) prediction, because they account for past engagements between users and ads. In this paper, we study how to efficiently construct historical features through counting features. The key challenge of such problem lies in how to automatically identify counting keys. We propose a tree-based method for counting key selection. The intuition is that a decision tree naturally provides various combinations of features, which could be used as counting key candidate. In order to select personalized counting features, we train one decision tree model per user, and the counting keys are selected across different users with a frequency-based importance measure. To validate the effectiveness of proposed solution, we conduct large scale experiments on Twitter video advertising data. In both online learning and offline training settings, the automatically identified counting features outperform the manually curated counting features.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
03/18/2020

Counting Bounded Tree Depth Homomorphisms

We prove that graphs G, G' satisfy the same sentences of first-order log...
research
10/02/2020

Interactive Reinforcement Learning for Feature Selection with Decision Tree in the Loop

We study the problem of balancing effectiveness and efficiency in automa...
research
02/25/2019

Entity Personalized Talent Search Models with Tree Interaction Features

Talent Search systems aim to recommend potential candidates who are a go...
research
08/11/2021

On the Explanatory Power of Decision Trees

Decision trees have long been recognized as models of choice in sensitiv...
research
07/25/2023

Feature Importance Measurement based on Decision Tree Sampling

Random forest is effective for prediction tasks but the randomness of tr...
research
03/04/2022

MF-Hovernet: An Extension of Hovernet for Colon Nuclei Identification and Counting (CoNiC) Challenge

Nuclei Identification and Counting is the most important morphological f...
research
12/09/2020

Counting methods introduced into the bibliometric research literature 1970-2018: A review

The present review of bibliometric counting methods investigates 1) the ...

Please sign up or login with your details

Forgot password? Click here to reset