Playing with hot spot and outlier analyses in Open Street Map

It's only recently that I get to use hot spot analyses beyond usual work. I just realized I never really used mapping for leisure or just about anything that I find interesting.

Now I plan to create more maps beyond my usual office stuff.

Here are some maps that I find interesting and might need further analyses.

SCHOOLS:

I generated a hot spot map of all the schools in the Philippines using the recent Open Street Map data. Basically a hot spot counts the number of instances there is a school within the bounding polygon set for analysis. In this case, the bounding polygon I used are cities/municipalities.

I then generated an optimized hot spot analysis that basically checks which cities has neighboring cities with higher values or lower values as themselves. It uses the Getis Ord-Gi* statistic to generate probability or p-values to quantify clustering. These clusters of high values then generates a hot spot and cluster of low values generate a cold spot. 

The darker red implies higher chances that this clustering among the neighboring cities is not a result of random chance hence the clustering is significant. 

Cold spots on the other hand or the the blue ones are cities containing very low cluster of schools as compared to its neighboring cities. Cagayan region on the North and Marawi area on the south are classified as cold spots. 

Light yellow on the other hand shows no significant clustering or neither hot or cold spot of schools in the Philippines. 

It's interesting to see the proliferation of schools being built beyond the greater manila area and central Luzon. It was however a question to me on why do we have cold spots in Ilocos and Cagayan region? Access or topography might be a factor but it's worth another spatial analysis.

Panay region is quite a surprise instead of the province of Cebu being generated as hot spot.

Davao in Mindanao is a given. Marawi is also a given cold spot.

Mapping is addicting. The more you map, the more you want to dig deeper into a phenomena.

I have generated another interesting map on outlier analytics of such generated hot spots. Will feature it in my succeeding posts.