联通手机信令统计大数据

利用个体位置信息还原个体在时间及空间纬度上的出行链路信息，探寻人群移动的特征规律。联通大数据的优势：中国联通是三家运营商第一个做到全国4G数据一点集中，数据格式全国统一。先进的手机信令数据处理技术，取代传统基站经纬度算法，精度提升到250m*250m的网格。成熟的数据质量管控和回溯机制，保证数据质量稳定、准确。与上百家的甲级规划院、高校和政府单位合作，形成行业模型积累和沉淀，辅助项目实施。依据联通大数据反推全量数据的算法：1.运营商数据扩样：采用联通市县一级的市场占比，同时参考工信部每人拥有号码个数，建立扩样算法公式。2.互联网数据扩样：采用互联网数据，进一步精确街道、网格级微小范围内联通市场占比，建立扩样算法公式。3.机器学习扩样：以联通用户为种子，采用卷积神经网络算法和深度学习将联通用户位置、通话时间、时长因素与对象位置建立联系，建立扩样算法公式。

This dataset restores the spatio-temporal travel trajectory information of individuals using their location data, aiming to explore the characteristic patterns of human mobility.

Advantages of China Unicom Big Data: China Unicom was the first among the three major domestic telecom operators to realize one-point centralized collection of national 4G data, with unified national data formats. Advanced mobile signaling data processing technology replaces traditional base station longitude-latitude algorithms, boosting positioning accuracy to 250m×250m grid cells. Mature data quality control and backtracking mechanisms ensure stable and accurate data quality. It has cooperated with hundreds of Class-A planning institutes, universities and government agencies, enabling the accumulation and refinement of industry-specific models to support project implementation.

Algorithms for inferring full-scale data based on China Unicom big data:
1. Telecom operator data sampling expansion: Establish a sampling expansion algorithm formula by adopting the market share of China Unicom at the city and county level, and referring to the number of mobile numbers per person issued by the Ministry of Industry and Information Technology (MIIT).
2. Internet data sampling expansion: Establish a sampling expansion algorithm formula by using internet data to further refine the market share of China Unicom at the street and grid-level micro regions.
3. Machine learning-based sampling expansion: Establish a sampling expansion algorithm formula by taking China Unicom users as seed samples, using Convolutional Neural Network (CNN) and deep learning to establish correlations between the location, call time and duration of China Unicom users and the locations of target users.

搜集汇总

背景与挑战

背景概述

该数据集由中国联通烟台分公司提供，采用4G信令数据实现250米精度的网格化位置分析，通过三种扩样算法还原人群移动特征。主要应用于政府规划、商业选址和金融风控等领域，需获得被查询对象授权使用。