Analytical Methodology

This section covers the statistical and machine-learning methodology behind AdvancedAnalytics: for every technique, a condensed R procedure immediately followed by the real AdvancedAnalytics tool, service or model-training family that does the equivalent job in the product today, or a plain statement that no equivalent exists and the procedure stays R-only.

R has not been erased from Jube’s analytics story. It is a reference point for understanding the product, and a genuine fallback for the residual cases the product does not cover, rather than the only way to do the work.

Not covered here

• Exhaustive Search. This is Jube’s own long-standing model family, not an R technique, and it already has thorough, current documentation: the as-built methodology specification at Exhaustive Methodology. Nothing here duplicates it.
• Deep Learning with H2O. H2O and deep learning are out of scope for AdvancedAnalytics entirely. There is no page for it in this section.

In this section

• Methodology and Taxonomy
• Introduction
• Getting Started with R
• R Server
• Data Structures
• Loading, Shaping and Merging Data
• Data Frames vs the Packed Array
• dplyr Gap Analysis
• Summary Statistics
• Variable Transformations
• Abstraction
• Charting
• Linear Regression
• Logistic Regression
• Decision Trees and Boosting
• Naive Bayes
• Neural Networks
• Bayesian Analysis with bnlearn
• Bayesian Networks
• Other Model Types
• Prescriptive Analytics
• Governance, KPIs and Model Validation
• Reference