SOA Exam PA (Predictive Analytics) – Practice Exams
About This Program
This program prepares candidates for the Predictive Analytics (PA) Examination of the Society of Actuaries (SOA), built on the SOA Exam PA syllabus. The real Exam PA is a three hour and thirty minute written-response examination: candidates analyze data in the context of a business problem and submit written responses to specified tasks in Microsoft Word, working with Word and Excel at a Prometric workstation. It contains no multiple-choice questions, and since April 2023 R and RStudio have not been available at the examination – all code and output relevant to the tasks is provided in the exam materials. This practice bank therefore works differently by design: its 1,500 four-option questions drill the concepts, model judgments, output interpretations, calculations, and communication choices that the written tasks demand, so the knowledge underneath your written answers is fast and reliable. The SOA does not publish a fixed passing percentage or raw cut score: Exam PA is graded in writing by SOA committees and results are reported on a 0-10 grade scale with 6 the passing grade, arriving about six weeks after the sitting under the SOA’s expedited-grading pilot. The SOA determines registration and reports results; Exam PA is an associateship actuarial examination, not a state license. This product is examination preparation only.
What Is Covered
- Generalized Linear Models (450 questions): selecting and validating a GLM for a business problem, offsets and weights, interpreting coefficients and interaction terms, and regularized regression with its hyperparameters.
- Data Exploration and Visualization (375 questions): data types, variables and data design, principles of constructing graphs, and univariate and bivariate exploration.
- Predictive Analytics Problem Definition (225 questions): descriptive, predictive and prescriptive analytics, the characteristics of predictive modeling problems, framing a vague business question into an analyzable problem, and feasibility and next steps.
- Data Transformations and Unsupervised Learning (225 questions): creating features that add value, principal components analysis, and K-means and hierarchical clustering used to transform data.
- Tree-Based Models (225 questions): constructing, pruning and validating regression and classification trees, and bagging, random forests and boosting.
How It Works
The bank contains 1,500 original practice questions, each with a full explanation, spanning the five syllabus topics. A timed knowledge simulation of 70 questions in 210 minutes mirrors the real sitting’s length and topic weights – it is a knowledge check, not a replica of the written task format, and your remaining preparation should include writing full task responses under time pressure. This is an independent study product: PassProfessor is not the SOA, does not administer Exam PA, and does not issue credentials.
