Data for: The Big Picture of Cities: Analysing Flickr Photos of 222 Cities Worldwide Using Machine Learning

The scope of the dataset is 222 cities in the Mercer’s quality of living (QoL) ranking 2019. The photos were retrieved from Flickr.com. To detect the objects and features of every photo, the authors used Google Cloud Vision, which is a recent technology that collects, analyses, and extracts information from visual images. All photos of the 222 cities were processed to detect their labels. The authors conducted latent Dirichlet allocation (LDA) modelling, which is the most common feature extraction or topic modelling algorithm in machine learning. Each photo was assigned the city image dimension including cityscape, landscape, architecture, transport, and recreation.

The table shows (1) the identification number of the city ordered by the Mercer’s QoL Ranking (2) the name of the photo file from 1-1,000, (3) the concatenated label of each photo, (4) the number of labels, (5) the Flickr ID of the photo, (6) the owner ID, (7) the type of license, (8) the date that the photo was taken, (9) view count, (10) latitude, (11) longitude, (12) LDA results, (13) the city image dimension.

本数据集的范围涵盖了2019年Mercer生活质量（QoL）排名中的222个城市。图片来源于Flickr.com。为了检测每张照片中的物体和特征，作者采用了谷歌云视觉技术，这是一种能够收集、分析和提取视觉图像信息的新兴技术。222个城市的所有照片均经过处理以检测其标签。作者进行了潜在狄利克雷分配（LDA）建模，这是机器学习中应用最广泛的特征提取或主题建模算法。每张照片都被赋予了城市图像维度，包括城市景观、自然风光、建筑、交通和娱乐等方面。

表格展示了以下内容：（1）按Mercer生活质量排名排列的城市识别号（2）照片文件的名称，从1-1,000号（3）每张照片的标签连接（4）标签数量（5）照片的Flickr ID（6）所有者ID（7）许可类型（8）照片拍摄日期（9）查看次数（10）纬度（11）经度（12）LDA结果（13）城市图像维度。