In nicolewhite/RNeo4j: Neo4j Driver for R

RNeo4j

RNeo4j is Neo4j's R driver. It allows you to read and write data from / to Neo4j directly from your R environment.

Contents

• Install
• Connect
• Nodes
• Relationships
• Cypher
• Shortest Paths
• Weighted Shortest Paths
• Graph Algorithms
• Visualizations
• Import
• Connection Issues
• Contribute

Install

Neo4j

First and foremost, download Neo4j!

Windows

If you're on Windows, download the .exe and follow the instructions. You'll get a GUI where you simply press "Start" to start Neo4j.

OS X

If you're on OS X, you can download either the .dmg or the .tar.gz. The .dmg will give you a GUI where you simply press "Start" to start Neo4j. Otherwise, download the .tar.gz, unzip, navigate to the directory and execute ./bin/neo4j start.

Linux

If you're on Linux, you have to use the .tar.gz. Download the .tar.gz, unzip, navigate to the directory and execute ./bin/neo4j start.

Bolt dependencies

These depencies are are only required if you want to use the Bolt interface. They must be present at build time, and libneo4j-client must also be present at runtime.

Windows

• Rust: https://rustup.rs
• Clang: binary releases are available at http://releases.llvm.org/download.html
• libneo4j-client:
  • Make sure you have RTools installed (necessary for building R packages on Windows)
  • Open your MinGW shell (check in start menu, C:\RTools\mingw_64, C:\RTools\mingw_32, C:\MinGW\bin, and C:\MinGW\1.0\bin)
  • See "Installing libneo4j-from source" section

OS X (with Homebrew installed)

• Rust: brew install rust (or https://rustup.rs but see "Rust Path" section)
• Clang: brew install llvm
• libneo4j-client: brew install cleishm/neo4j/libneo4j-client

Linux

• Rust:
  • Debian based (e.g. Mint, Ubuntu): sudo apt-get install cargo
  • Arch Linux: sudo pacman -S rust
  • Building from source: https://rustup.rs but see "Rust Path" section
• Clang: get it from your package manager
  • Debian based (e.g. Mint, Ubuntu): sudo apt-get install clang libclang-dev
  • Arch Linux: sudo pacman -S clang
  • Other: your package manager almost certainly has clang. It may be called llvm.
• libneo4j-client:
  • Debian based: sudo apt-get install libneo4j-client-dev
  • Other: See "Installing libneo4j-client from source"

Rust Path

By default, on *nix systems (such as Linux and OS X), rustup only sets the PATH in your shell. That means that if you try to build RNeo4j in a GUI application like RStudio, it may fail. To work around this issue, simply build RNeo4j in a terminal.

RNeo4j

From CRAN

install.packages("RNeo4j")

From GitHub

devtools::install_github("nicolewhite/RNeo4j")

From Source

Go to the latest release and download the source code. You can then install with install.packages.

install.packages("/path/to/file.tar.gz", repos=NULL, type="source")

Load the Package

library(RNeo4j)

Connect

graph = startGraph("http://localhost:7474/db/data/")

If you have authentication enabled, pass your username and password.

graph = startGraph("http://localhost:7474/db/data/", username="neo4j", password="password")

clear(graph, input=F)

Nodes

nicole = createNode(graph, "Person", name="Nicole", age=24)
greta = createNode(graph, "Person", name="Greta", age=24)
kenny = createNode(graph, "Person", name="Kenny", age=27)
shannon = createNode(graph, "Person", name="Shannon", age=23)

Relationships

r1 = createRel(greta, "LIKES", nicole, weight=7)
r2 = createRel(nicole, "LIKES", kenny, weight=1)
r3 = createRel(kenny, "LIKES", shannon, weight=3)
r4 = createRel(nicole, "LIKES", shannon, weight=5)

Cypher

If you're returning tabular results, use cypher, which will give you a data.frame.

query = "
MATCH (nicole:Person)-[r:LIKES]->(p:Person)
WHERE nicole.name = 'Nicole'
RETURN nicole.name, r.weight, p.name
"

cypher(graph, query)

For anything more complicated, use cypherToList, which will give you a list.

query = "
MATCH (nicole:Person)-[:LIKES]->(p:Person)
WHERE nicole.name = 'Nicole'
RETURN nicole, COLLECT(p.name) AS friends
"

cypherToList(graph, query)

Both cypher and cypherToList accept parameters. These parameters can be passed individually or as a list.

query = "
MATCH (p1:Person)-[r:LIKES]->(p2:Person)
WHERE p1.name = {name1} AND p2.name = {name2}
RETURN p1.name, r.weight, p2.name
"

cypher(graph, query, name1="Nicole", name2="Shannon")
cypher(graph, query, list(name1="Nicole", name2="Shannon"))

Shortest Paths

p = shortestPath(greta, "LIKES", shannon, max_depth=4)
n = nodes(p)
sapply(n, "[[", "name")

Weighted Shortest Paths

p = shortestPath(greta, "LIKES", shannon, max_depth=4, cost_property="weight")
n = nodes(p)
sapply(n, "[[", "name")
p$weight

Graph Algorithms

library(igraph)

query = "
MATCH (n)-->(m)
RETURN n.name, m.name
"

edgelist = cypher(graph, query)
ig = graph.data.frame(edgelist, directed=F)

betweenness(ig)
closeness(ig)

Visualizations

igraph

plot(ig)

ggnet

library(network)
library(GGally)

net = network(edgelist)
ggnet(net, label.nodes=TRUE)

visNetwork

Read this blog post and check out this slide deck.

Import

clear(graph, input=F)

library(hflights)
hflights = hflights[sample(nrow(hflights), 1000), ]
row.names(hflights) = NULL

head(hflights)

addConstraint(graph, "Carrier", "name")
addConstraint(graph, "Airport", "name")

query = "
CREATE (flight:Flight {number: {FlightNum} })
SET flight.year = TOINT({Year}),
    flight.month = TOINT({DayofMonth}),
    flight.day = TOINT({DayOfWeek})

MERGE (carrier:Carrier {name: {UniqueCarrier} })
CREATE (flight)-[:OPERATED_BY]->(carrier)

MERGE (origin:Airport {name: {Origin} })
MERGE (dest:Airport {name: {Dest} })

CREATE (flight)-[o:ORIGIN]->(origin)
CREATE (flight)-[d:DESTINATION]->(dest)

SET o.delay = TOINT({DepDelay}),
    o.taxi_time = TOINT({TaxiOut})

SET d.delay = TOINT({ArrDelay}),
    d.taxi_time = TOINT({TaxiIn})
"

tx = newTransaction(graph)

for(i in 1:nrow(hflights)) {
  row = hflights[i, ]

  appendCypher(tx, query,
               FlightNum=row$FlightNum,
               Year=row$Year,
               DayofMonth=row$DayofMonth,
               DayOfWeek=row$DayOfWeek,
               UniqueCarrier=row$UniqueCarrier,
               Origin=row$Origin,
               Dest=row$Dest,
               DepDelay=row$DepDelay,
               TaxiOut=row$TaxiOut,
               ArrDelay=row$ArrDelay,
               TaxiIn=row$TaxiIn)
}

commit(tx)

summary(graph)

Connection Issues

Couldn't connect to server

Error in curl::curl_fetch_memory(url, handle = handle) : 
  Couldn't connect to server

Neo4j probably isn't running. Make sure Neo4j is running first. It's also possible you have localhost resolution issues; try connecting to http://127.0.0.1:7474/db/data/ instead.

No authorization header supplied

Error: client error: (401) Unauthorized
Neo.ClientError.Security.AuthorizationFailed
No authorization header supplied.

You have auth enabled on Neo4j and either didn't provide your username and password or they were invalid. You can pass a username and password to startGraph.

graph = startGraph("http://localhost:7474/db/data/", username="neo4j", password="password")

You can also disable auth by editing the following line in conf/neo4j-server.properties.

# Require (or disable the requirement of) auth to access Neo4j
dbms.security.auth_enabled=false

Contribute

Check out the contributing doc if you'd like to contribute!

nicolewhite/RNeo4j documentation built on May 23, 2019, 5:09 p.m.