Resources by admin

Relative Wealth Index Data

Relative Wealth Index Data

admin
Open Dataset
Publication Date: 1 January 2021
Creators: Chi, Guanghua; Fang, Han; Chatterjee, Sourav; Blumenstock, Joshua E.

The Relative Wealth Index predicts the relative standard of living within countries using de-identified connectivity data, satellite imagery and other nontraditional data sources. It has been built by researchers at the University of Carlifornia – Berkeley and Facebook. The estimates are built by applying machine learning algorithms to vast and heterogeneous data from satellites, mobile phone networks, topographic maps, as well as aggregated and de-identified connectivity data from Facebook. They train and calibrate the estimates using nationally-representative household survey 20 data from 56 LMICs, then validate their accuracy using four independent sources of household survey data from 18 countries. They also provide confidence intervals for each micro-estimate to facilitate responsible downstream use. The data is provided for 93 low and middle-income countries at 2.4km resolution. It covers the time between April 01, 2021 and December 22, 2023.

An interactive map of the Relative Wealth Index is available here: http://beta.povertymaps.net/

A Novel Integrated Network with LightGBM for Click-Through Rate Prediction

A Novel Integrated Network with LightGBM for Click-Through Rate Prediction

admin
Journal Article
Publication Date: 13 October 2021
Creators: Xia, Zhen; Mao, Senlin; Bai, Jing; Geng, Xinyu; Yi, Liu

Click-through Rate (CTR) prediction has become one of the core tasks of the recommendation system and its online advertising with the development of e-commerce. In the CTR prediction field, different features extraction schemes are used to mine the user click behavior to achieve the maximum CTR, which helps the advertisers maximize their profits. At present, achievements have been made in CTR prediction based on Deep Neural Network (DNN), but insufficiently, DNN can only learn high-order features combination. In this paper, Product & Cross supported Stacking Network with LightGBM (PCSNL) is proposed for CTR prediction to solve such problems. Firstly, the L 1 and L 2 regularizations are imposed on Light Gradient Boosting Machine (LightGBM) to prevent overfitting. Secondly, the method of vector-wise feature interactions is added to product layer in product network to learn second-order feature combinations. Lastly, feature information is fully learned through the cross network, product network and stacking network in PCSNL. The online ads CTR prediction datasets released by Huawei and Avazu on the Kaggle platform are involved for experiments. It is shown that the PCSN model and PCSNL have better performance than the traditional CTR prediction models and deep learning models.

Sign In

New to BERD@NFDI? Register

Forgot password?

Register

Reset Password

Please enter your username or email address, you will receive a link to create a new password via email.