- tell you my story
- maybe inspire you to program
- a brief lesson on spatial analysis
- anything you want this is your time

or, why should you actually listen to me

- MS in Urban Informatics, Northeastern
- Sr. Product Engineer @ Esri
- Spatial Statistics

- Previously @ RStudio (now Posit)
- Programming in R since 2014

but not in the way that most people want to be

“…this New Hampshire institution of higher learning is back. Four things students dig are skiing, skiing, studying and [partying] on the lake.

from *anthropology* & *sociology* to GIS

“everything is related to everything else, but near things are more related than distant things.”

everything we do happens *somewhere*

- wanted to do more advanced stats
- learned 1:1 with my professor
- finding data meant cleaning data
- explored new packages and exposed to new domains
- natural language
- software engineering
- etc

- spend all my free time learning R
- I intern at DataCamp making R courses
- Research using interactive GIS
- Decide grad school was the right thing for me
- promptly rejected from all but one school

- one of my last courses in grad school
- learned about spatial autocorrelation
- spatial regression
- basics of networks analysis

neighborhoods, autocorrelation, and tooling

fundamental concept of analysis ## In urban studies

- the neighborhood fundamental to sociology
- Chicago school (Park & Burgess)

- used to understand differences inside of the city
- too much nuance

- are phenomena spatially dependent?
- do similar values occur near each other

- start focal with a location \(i\)
- it’s neighbors are \(j\)
- \(X_i\) is compared to \(X_j\)
*not*to \(X\)

it depends….

how do you choose what the neighbors are for a location?

how does \(i\) compare to \(j\)

the neighborhood value

- “expected value” of the neighborhood
- it is the average value of the neighborhood (excluding \(i\))
- summarizes values of \(x\) for an observation \(i\)’s neighborhood

- spatial clustering (autocorrelation)
- hot spot detection (clustering)
- spatial regression
- (inference / neighborhood spill over effects)

- spatio-temporal hot spot analysis

`sf`

- spatial vector data`spdep`

- spatial statistics`sfdep`

- a tidy interface to spdep`rgeoda`

- R interface to GeoDa

`geopandas`

-`sf`

equivalent`pysal`

a very robust set of spatial statistics toools`shapely`

for geometries

- i can answer questions
- i can demo code
- i can talk through the spatial lag in more detail
- we can discuss hot spot analysis
- anything you want :)