An Introduction to the Field for Political Operatives

Data Analytics

What advantage can the political operative gain by the use of data analytics? To answer this question we must first inventory the concerns of the operative through the lens of a typical campaign cycle. These concerns include:

  1. Defining the position of the candidate. What are the important issues that drive voter action? What levers or triggers cause a measurable activity to occur?
  2. The mother’s milk issue is, of course, what activities result in monies flowing to the candidate or cause. The first item above speaks to the causative levers of contribution but it is an incomplete picture. In this area form follows function and strongly influences the efficiency and effectiveness of the contribution collection mechanism. How do we gain an understanding of the actions that have resulted in greater gains and how can we extrapolate our result?
  3. Given the issues base of the candidate, what are the most likely sources of contribution? To what networks of influence do these sources belong and may they be fruitful as potential donors?
  4. What should my reaction to the polls be? Do they provide an opportunity for immediate action?
  5. What are the most effective means of advertising? How should each be used to capitalize on positives or to offset negatives?
  6. What is the most effective means to reinforce the perception of candidate character? What are the benefit/costs of such efforts? What unintended consequences should I monitor for when we undertake a character development initiative?
  7. What are the most productive influence networks and how are they best accessed?
  8. What endorsements are of value and which may cause offsetting value?
  9. Given the truism that the media is the message, what media types and channel combinations offer an optimal result?

Each of these questions may be informed by the use of data analytics. Data analytics provides tools, as generalized categories, to:

I. Discover relationships between entities – These methods allow us to discover the network of relationships that may exist answering question three, seven and to some extent nine above. More than one method may be used from the segmentation and clustering and / or the social network analysis sets.

II. Predict or anticipate future events – Again we may use a mix of methods, from the predictive modeling field we can estimate the level of contributions that may be expected to help answer question two, arguably the single most important question of the set. The use of recommender systems may also help us answer questions one, five and nine.

III. Change event analysis and reaction – All of the data analytic methods approaches discussed below may add to our ability to understand and react to changes in the environment or event stream. We can look to the techniques of sentiment mining, text mining and social network analysis to help us answer the question posed in items four, six and eight above.

The naming and definitions of the techniques are methods are arbitrary but have begun to reach a normative state. The definitions below are redacted from the website and will be used through the balance of this data analytics series to discuss the detail of each method.

Predictive Modeling and Forecasting

In predictive modeling (also called predictive analytics) we seek to predict the value of a variable of interest (purchase/no purchase, fraudulent/not fraudulent, malignant/benign, amount of spending, etc.) by using “training” data where the value of this variable is known.  Once a statistical model is built with the training data (“trained”), it is then applied to data where the value is unknown.  Predictive modeling is also termed “supervised learning” and is covered in the following courses:

Recommender Systems

The purpose of a recommender system is to identify, statistically, “what goes with what.”  These systems lie behind the notices you see on web sites advising you that “customers who bought X also bought Y.”  The general statistical terms for the methods used are affinity analysis and association rules; these are unsupervised methods.

  • Data Mining – Unsupervised Methods
  • Decision Trees


In clustering, we seek to identify groups of customers, records, etc. that are similar to one another.  “Clustering” is the general statistical technique; when we apply it to customers it is the statistical component in customer segmentation.  Clustering is an “unsupervised” data mining method – there is no known outcome that serves to train a model.

  • Cluster Analysis
  • Data Mining – Unsupervised Methods

Text Analytics & Social Network Analysis

The most rapid data growth is not in numerical data, but in text – Twitter feeds, the contents of Facebook pages, emails, etc. – which must be pre-processed to be usable.  

  • Text Mining
  • Natural Language Processing
  • Sentiment Analysis
  • Social Network Analysis ( LLC, 2004-2013)

Works Cited LLC. (2004-2013). Data Analytics. Retrieved from

James Strawn About James Strawn

James is an IT consultant and solutions architect with decades of experience working on various major projects with Fortune 500 companies. He is also an author and editor for, a blog forum for politically active citizens and professionals.

Speak Your Mind