Visualizations
This Page presents visual insights of data, artifacts, and communications
Email Archives
Figure 1:
Figure 1 is a visualization of communication networks within Enron based on connections in an individuals network of contacts (addresses sent and received), as well as the number of connections between contacts within that individuals network. Large nodes indicate that that individual is member or center to a significantly interconnected communication network.
Figure 2:
Figure two takes the same variable of interconnectedness, but applied to a subset of data sorted by job level. data set was sorted by grouping data labeled with job titles into 9 categories. Model measures average connections of different organizational roles. However, this model is likely unreliable as the majority of the dataset falls under 'NA'. Though there is a clear trend of a smaller communication network as position within the company increases.
Figures 3 & 4:
Figure 1:
Figure 2:
Figure 3 utilizes a open source sentiment analysis function 'vaderSentiment,' which analyzes vocabulary, punctuations and other aspects of text and assigns a value of positive or negative. Results show email communication between this group of employees were overwhelmingly positive in nature. My findings are confirmed by figure 4, taken from a study by Dr. David Noever on using machine learning to find indications of fraud. According to Noever, "Given the dire cloud surrounding its eventual downfall, the email corpus shows more positive sentiment than one might initially expect. Trust and anticipation dominate fear, disgust, sadness and anger."
Figure 3:
Figure 4:
