In matthiasronnau/graphdata: Put Purchase Data Into Graph Format

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(graphdata)

Generating graph data

The package "graphdata" is a simple package that transforms purchase data into an iGraph object. The iGraph package is useful for visualizing and computing various statistics about a social network, but sometime data must be transformed in order to work properly. This package takes event level data and transforms it into a network that is workable with iGraph.

Generating the `iGraph` object

The dataframe must have a unique identifier to group by, such as a user id. It must also have a unique identifier for every product in the network, such as a product id.

set.seed(277)
df <- data.frame("user_identifier" = rep(1:3, 3), 
                 "product" = 277:285, 
                 "price" = c(sample(1:500, 4, replace = TRUE), NA, 
                             sample(1:500, 4, replace = TRUE)))

The data to be used must be cleansed of all missing values.

cleaned_df <- na.omit(df)

The cleaned data can now be transformed into an iGraph object with graph_data().

g <- graph_data(data = cleaned_df, id = "user_identifier", product_id = "product")
g

The resulting object can now be visualized as a network or transformed into a dataframe using built in iGraph functions in order to compute various statistics about the network.

Using graph_data in conjucntion with igraph

Once the graph object has been generated with the graph_data(), it is possible to use it in conjunction with the various functions available in igraph, such as visualizing the network.