Joint Research Meeting

ISBIS StatFin 2026

7-11 December, 2026

Workshop

The Joint 2026 Annual Meeting of the International Society for Business and Industrial Statistics (ISBIS) and the Eleventh Meeting on Statistical Methods in Finance (StatFin) are pleased to announce three intensive pre-conference workshops.

These technical short courses are explicitly designed to bridge theoretical foundational principles with cutting-edge industry and public sector applications.

Date and Venue 7-8 Dec, 2026 Chennai Mathematical Institute

Target Audience

This program is curated for:

• PhD Students & Researchers looking to enhance their practical methodological skill set.

• Young Faculty Members seeking contemporary research tools and teaching frameworks.

• Young Professionals in Government (such as the Indian Statistical Service (ISS) and Indian Economic Service (IES)) managing complex data environments and policy modeling.

• Industry Professionals applying data science, econometrics, and language models in corporate ecosystems.

Workshop 1: How to Train your Language Models

Workshop 2: Asset Pricing Meets Econometrics: Classical and Bayesian Regression in Practice

Workshop 3: The What, Why and How of Court Data

Workshop 1: How to Train your Language Models

• Instructors: Sudhir Kumar (Coriolis Technologies) & Sourish Das (Chennai Mathematical Institute)
• Focus: An operational and statistical deep-dive into the mechanics of Large Language Models (LLMs). This course covers the foundations of training architectures, optimization challenges, and fine-tuning methodologies.

Course Outline:

• Module 1: Foundations of Language Models An introduction to language modeling, exploratory data analysis (EDA) of textual corpuses, and data cleaning strategies.
• Module 2: Statistical Mechanics of Tokenization Understanding the mathematical and statistical underpinnings of BPE (Byte Pair Encoding) tokenizers.
• Module 3: Word Embeddings and Vector Spaces Exploring word representations, Word2Vec, semantic geometries, and vector space logic.
• Module 4: Sequence-to-Sequence Modeling and Attention Mechanics From Recurrent Neural Networks (RNNs) to Self-Attention mechanisms and the core architecture of Transformers.

Hands-on Lab Sessions (using Google Colab):

Participants will gain hands-on implementation experience using a suite of production-ready interactive notebooks:

• Notebook 1: Data Pipeline Build an end-to-end PDF text extraction, document chunking, and metadata annotation pipeline.
• Notebook 2: SLM Fine-Tuning Conduct a practical Low-Rank Adaptation (LoRA) fine-tuning experiment on the Phi-3.5-mini-instruct Small Language Model.
• Notebook 3: SLM Few-Shot Learning Build and evaluate an 8-shot document classifier complete with rigorous statistical evaluation and error analysis.
• Notebook 4: Interactive Classifier Demo Deploy a live browser-based interface for your document classifier using the Gradio framework.
• Notebook 5: Conversational RAG Engine Implement a conversational Retrieval-Augmented Generation (RAG) chatbot capable of contextual reasoning over your custom document corpus.

Workshop 2: Asset Pricing Meets Econometrics: Classical and Bayesian Regression in Practice

• Instructors:: By M.A. Rahman, (IIT Kanpur)

• Focus: This workshop provides a structured introduction to classical and Bayesian approaches to linear regression, with applications to financial econometrics. Using stock price data for Amazon, participants will learn to estimate and interpret the Capital Asset Pricing Model (CAPM) and the Fama–French three-factor model under both frameworks. The workshop is designed for advanced undergraduate students, graduate students and researchers in economics, finance, and statistics.

1. Foundations and Classical Linear Regression
2. Bayesian Linear Regression in Depth

By the end of the workshop, participants will understand the conceptual and practical differences between classical and Bayesian regression methods, and will be able to implement both approaches in empirical settings. The workshop consists of four lectures of approximately 1.5-2 hours each, combining theoretical exposition with applied illustration. Supporting materials, including datasets and code templates, will be provided.

Prerequisite Participants are expected to have a basic background in probability, statistics, and linear regression.

Workshop 3: The What, Why and How of Court Data

• Instructor: Bhargavi Zaveri-Shah (The Professor)

• Focus: An essential guide to extracting, processing, and analysing legal and administrative datasets. Perfect for public sector economists, policy analysts, and data scientists interested in empirical legal studies, governance metrics, and quantitative institutional evaluation.

A goldmine data sets for Statisticians and Data Scientists.