期刊
JOURNAL OF OFFICIAL STATISTICS
卷 39, 期 4, 页码 535-570出版社
SCIENDO
DOI: 10.2478/jos-2023-0025
关键词
Big data; High frequency statistics; dynamic population mapping; spatial accuracy
This article proposes a methodology to estimate the hourly population presence over France consistently over several months based on spatially merged fine-grained official population counts and signaling data. The article highlights the importance of consistency at different spatial scales and over time, as well as spatial mapping reflecting accuracy. The challenges and potential of data fusion approaches for future methodologies are also discussed.
Mobile network data records are promising for measuring temporal changes in present populations. This promise has been boosted since high-frequency passively-collected signaling data became available. Its temporal event rate is considerably higher than that of Call Detail Records - on which most of the previous literature is based. Yet, we show it remains a challenge to produce statistics consistent over time, robust to changes in the measuring instruments and conveying spatial uncertainty to the end user. In this article, we propose a methodology to estimate - consistently over several months - hourly population presence over France based on signaling data spatially merged with fine-grained official population counts. We draw particular attention to consistency at several spatial scales and over time and to spatial mapping reflecting spatial accuracy. We compare the results with external references and discuss the challenges which remain. We argue data fusion approaches between fine-grained official statistics data sets and mobile network data, spatially merged to preserve privacy, are promising for future methodologies.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据