Intro to Graph Databases (with R)

Bozeman R useR Group

Tidy Data

csv files

# populations.csv

state          , population
California     , 38332521
Texas          , 26448193
New York       , 19651127
Florida        , 19552860
Illinois       , 12882135
Pennsylvania   , 12773801
Ohio           , 11570808
Georgia        , 9992167
Michigan       , 9895622
North Carolina , 9848060
New Jersey     , 8899339
...

tabular or relational databases (sql)

# populations.csv

state   , population , county                 , population
Montana , 1032949    , Yellowstone County     , 157048
Montana , 1032949    , Missoula County        , 114181
Montana , 1032949    , Gallatin County        , 100739
Montana , 1032949    , Flathead County        , 96165
Montana , 1032949    , Cascade County         , 82278
Montana , 1032949    , Lewis and Clark County , 66418
...
SELECT C.county, S.year_founded
FROM Counties C
JOIN States S on S.state = C.state

graph databases

Components of a graph

  • nodes
    • nodes have properties
  • relationships
    • relationships connect two nodes
    • relationships can have properties

Benefits of graph

  • Graph databases are efficient for highly connected data
  • Graph queries are often simpler than SQL counterparts
  • Model/schema is probably closer to truth with graph
  • Your thought process with graph will be different

How do we use graph?

  • Exploring and visualize the graph itself
  • Querying the graph for sub-graphs
  • Sub-graphs may be expressed as a table

Our first graph database

neo4j

neo4j: https://neo4j.com/

neo4j running: http://localhost:7474/

neo4j R driver: https://github.com/nicolewhite/RNeo4j

neo4j cypher documentation: https://neo4j.com/docs/developer-manual/current/cypher/

Example

Scenario

  • Record the ID’s of sheep when they come into contact

github.com/useRbozeman/intro-to-graphdb

data > contacts.csv

Question 1

  • Find me the sheep who have contacted the most other sheep?

Question 2

  • Find me the sheep who have have the most up-to-second order contacts?

Question 3

  • Find me the shortest contact path between sheep 1 and sheep 2

Question 4

  • Expansion: I want to weight the contacts by duration of contact.

  • Who has most total contact time?

github.com/useRbozeman/intro-to-graphdb

data > graph1.csv

Example

Scenario

A single contact event can include more than two sheep

github.com/useRbozeman/intro-to-graphdb

data > graph2.csv

Question 1

  • Which sheep have the largest average (in number of contacts) contact event?

Question 2

  • Which sheep have very little contact?

Example

Scenario

Contact with other things!

Question 1

  • What are all the things the sheep has had contact with?