Formable Advanced High-Strength Steels (AHSS) have a unique combination of strength and ductility, making them ideal in the effort to lightweight vehicles. The AHSS in this study, Quenched and Partitioned 1180, rely on the Transformation Induced Plasticity (TRIP) effect, in which retained austenite (RA) grains transform to martensite during plastic deformation, providing extra ductility via the transformation event. Understanding the factors involved in RA transformation, such as local strain and grain attributes, is therefore key to optimizing the microstructure of these steels. This research seeks to increase understanding of those attributes and the correlations between microstructure and RA transformation in TRIP steels. To measure local strain, the viability of using forescatter detector (FSD) images as the basis for DIC study is investigated. Standard FSD techniques, along with an integrated EBSD / FSD approach (Pattern Region of Interest Analysis System), are both analyzed. Simultaneous strain and microstructure maps are obtained for tensile deformation up to around 6% strain. The method does not give sub-grain resolution, and surface feature evolution prevents DIC analysis across large strain steps; however, the data is easy to obtain and provides a natural set of complementary information for the EBSD analysis. In-situ tensile tests combined with EBSD allow RA grain and neighboring attributes to be characterized and corresponding transformation data to be obtained. However, pseudo-symmetry of the ferrite (BCC) and martensite (BCT) phases prevents EBSD from accurately identifying all phases. Measuring the relative distortion of the crystal lattice, tetragonality, is one approach to identifying the phases. Unfortunately, small errors in the pattern center can cause significant errors in tetragonality measurement. Therefore, this research utilizes a new approach for accurate pattern center determination using a strain minimization routine and applies it to tetragonality maps for phase identification. Tetragonality maps based on dynamically simulated patterns result in the most accurate maps and can also be used to predict approximate local carbon content. Machine learning is then used on the collected data to isolate key attributes of RA grains and provide a decision tree model to predict transformation based on those attributes. Among the most relevant attributes found, RA grain area, RA grain shape aspect ratio, a “hardness” factor, and major axis orientation are included. Possible correlations between these factors and transformation improve understanding of relevant attributes and show the advantage that machine learning can have in unravelling complex material behavior.