Connect to the DB

A neo4j database needs to be running locally with data in it. We can connect to it in R via:

library(RNeo4j)

#data is already loaded into database
gdb <- startGraph("http://localhost:7474/db/data", user="neo4j", 
                  password = "maps") #only the password is DB specific here

Note the last comment. The default user is neo4j and the port and path are all defaults. Only the password is mine, and since this is a local DB with no sensitive info, I’m not concerned about revealing the password in a blog post, or trying to pick a secure password.

Basic Counting

We’ll start with some basic counting of courses and major structure. For each major, we list the number of courses required by that major. This count only includes specific courses required such as MAT-175, if the major requires either MAT-175 or MAT-275 that is an option group (or component) counted later.

query <- 'MATCH (c:Course)-[:Satisfies{type:"required"}]->(mc:Component{name:"Core"})-[:Part_Of]->(m:Major) 
RETURN m.name AS Major, count(c) AS NumberClasses, sum(toInt(c.minCredits)) AS CreditMin'
gq <- cypher(gdb,query) %>% as.data.frame()

kable(arrange(gq, Major, desc(NumberClasses)), caption = "Number of courses required for each major")

Next we look at the number of “option groups” or components for each major. These are the number of groups of courses from which a selection of credits or courses must be completed. We then look at the number of courses and minimum credits available for each option group in each major.

query <- 'MATCH (c:Component)-[p:Part_Of]->(m:Major) WHERE c.name<>"Core"
RETURN m.name AS Major, c.name AS OptName'
gq <- cypher(gdb,query) %>% as.data.frame()

gq <- group_by(gq, Major) %>% summarise(Number.Option.Groups = n_distinct(OptName))
kable(arrange(gq, Major, desc(Number.Option.Groups)), caption = "Number of Option Groups required for each major")

query <- 'MATCH (c:Course)-[s:Satisfies]->(comp:Component)-[p:Part_Of]->(m:Major) WHERE comp.name<>"Core" 
RETURN m.name AS Major, comp.name AS Component, count(c) AS NumberClasses, sum(toInt(c.minCredits)) AS CreditMin'

gq <- cypher(gdb,query) %>% as.data.frame()

kable(arrange(gq, Major, Component, desc(NumberClasses)), caption = "Number of courses for each option group of each major")

Major Visuals

Now let’s use graphs to visualize the majors in the MAPS department. I’m going to look at each major separately for now. What follows is the same function from the last post, except with “minor” replaced with “major”. VisNetork requires a dataframe of nodes and edges as input which we gather from separate queries - one for each type of node/edge.

library(visNetwork)
majorVis <- function(majorName){
  
  MajorNodeQ <- paste0('MATCH (m:Major {name:"',majorName,'"}) 
                       RETURN m.name AS id, m.name AS label, LABELS(m)[0] AS group')

  ComponentNodeQ <- paste0('MATCH (c:Component)-[:Part_Of]->(m:Major{name:"', majorName,'"}) 
                           RETURN c.name AS id, c.name AS label, LABELS(c)[0] AS group')

  CourseNodeQ <- paste0('MATCH (c:Course)-[:Satisfies]->(:Component)-[:Part_Of]->(m:Major{name:"',majorName,'"}) 
                        RETURN c.id AS id, c.id AS label, LABELS(c)[0] AS group')

  nodes <- rbind.data.frame(cypher(gdb, MajorNodeQ), 
                            cypher(gdb, CourseNodeQ))
  nodes <- rbind.data.frame(nodes, cypher(gdb, ComponentNodeQ))
  nodes <- nodes[!duplicated(nodes),]

  edgeSatQ <- paste0('MATCH (c:Course)-[r:Satisfies]->(co:Component)-[:Part_Of]->(m:Major {name:"',majorName,'"}) 
                     RETURN c.id AS from, co.name AS to, TYPE(r) AS label')

  edgePOQ <- paste0('MATCH (c:Component)-[r:Part_Of]->(m:Major {name:"',majorName,'"}) 
                    RETURN c.name AS from, m.name AS to, TYPE(r) AS label')

  edges <- rbind.data.frame(cypher(gdb, edgeSatQ),cypher(gdb, edgePOQ))

  visNetwork(nodes, edges)
}

Next we’ll call this function for each of the majors in MAPS. VisNetwork produces JavaScript graphs, so these may take some time to load. Also, you can drag nodes and edges around some (visNetwork bounces them back unless they move “enough”) as well as zoom in and out.