Week 5
Statistical Analysis of Urban Data
This week, we will discuss the unique features of geospatial data and the importance of considering the spatial structure of most urban data when conducting statistical analysis. We will then have our first introduction to individual-level GPS data, and will discuss different techniques to extract semantic meaning and quantitative behavioral measures from GPS observations.
In our first practical, we will build on our results from week 2 to create geodemographic classifications of urban areas using census data. In our second practical, we will transform raw individual-level GPS data into meaningful behavioral information: detecting stay-points, behavioral motifs, displacement lengths, and creating semantically-enriched mobility networks.
Prepare
đź“– Take a look at two contrasting approaches to characterizing urban behavior: (1) with spatial clustering of demographic variables from the census [1], and (2) using individual-level behavioral data [2]. Finally, re-visit a classic review of mobile phone data mining [3].
Creating the 2011 area classification for output areas (2011 OAC) [1]
- This article is a dense methodological piece! Please just look at sections 1, 2, 7, and 8.
- You can also explore the 2011 OAC: CDRC Mapmaker - 2011 Output Area Classification
Identifying latent activity behaviors and lifestyles using mobility data to describe urban dynamics [2]
- This article gives a good introduction to the behavioral insights that can come from GPS mobility data. Contrast these results with the approach taken in [1].
A review of Urban Computing for Mobile Phone Traces [3]
- This is a classic review of trajectory creation and mining which we saw in week 3. We will repeat much of the analysis mentioned in this paper in practical 2.
Participate
Perform
⌨️ Lab 5-1 – Statistical Analysis of Urban Data
Back to course schedule ⏎