Meridian 1.6.0: Enhanced EDA package and accelerated workflows with JAX

[kmm022]

New update (as of v1.6.0) – Meridian now offers a high-performance JAX backend and an enhanced Exploratory Data Analysis (EDA) package, empowering you to build more reliable models with significantly faster iteration speeds and deeper data transparency.

1. EDA Package – Meridian’s enhanced EDA package allows you to generate a comprehensive, HTML report to visualize and validate your data before you begin modeling. The report includes clear descriptions of potential data issues with recommended action items, providing a smoother path to resolve common data errors. Also, with new configurable thresholds, you can tailor these diagnostics to your specific dataset, ensuring your modeling foundation is solid and aligned with your business context. Export your EDA report to Google Drive to easily share with stakeholders for improved understanding of your model inputs.

2. JAX Backend – For projects that require enhanced performance and memory efficiency, accelerate your Meridian workflows by leveraging a JAX backend. Based on testing, leveraging JAX in Meridian streamlines model iterations, enabling 2x faster runtimes, 4x less memory usage, and uninterrupted workflows by eliminating kernel restarts. This increased efficiency allows you to leverage more computationally intensive parameters or enable 64-bit precision for improved stability in complex scenarios, ensuring your model remains robust and scalable without compromising the integrity of your ROI insights.

Important Note: By default, Meridian still uses TensorFlow for its core numerical operations and probabilistic modeling. However, in H2 2026, we will switch the default backend to JAX. This will be done within a major release, and we will notify you via our Meridian channels (GitHub Announcements, Discord) with an exact launch date as it approaches.

Get started today by generating your EDA report and seeing JAX in action with the Getting started with JAX notebook!