2. Hypothesis/EDA or Exploratory Data Analysis
● I didn’t have an hypothesis per se. My goals was rather to paint an nuanced
picture of the process of urbanization in China over the past 40 years
● At the core of the picture where 8 Satellite imagery datasets that I got from
the NASA Earth Observatory
● My goal was to explore the transformation of Shanghai over the past 40 years
3. Data Acquisition
I got my data from three sources:
1. Nasa Earth Observatory
2. Beijing City Lab
3. Atlas of Urban expansion
4.
5.
6. Analysis
● My first step was georeferencing each of the 8 rasters I downloaded
● The next step was performing ISO-cluster classification on each of the
images. I chose to produce three classes
● I then used map-algebra to compare the newly classified rasters
● This involved me subtracting one raster from another to produce difference
rasters
● This allowed me to highlight change between these rasters
11. Processed data:Iso Classification
How iso classification works, “ISO Classification is an unsupervised classification
method that separates that image into natural classes with the image. The ISO is
short for IsoData and stands for Iterative Self-Organizing Data Analysis
Technique. The clustering algorithm iteratively finds clusters based on minimum
distance techniques. In other words, it makes a set of clusters and if the standard
deviation within the cluster is too wide, in the next step it reclusters the items with
the algorithm. “
I this part of the project, I processed each of the 8 images into an IsoClassified
raster with 6 classes. The goal with iso classification was to highlight certain
features of the raster.