LLMs IN DATA PREPARATION AND MODELING
SUPPORT
Introduction to LLMs in Predictive
Workflows
Role of LLMs in the data science lifecycle
Complementing vs. replacing traditional ML tools
Tools overview: ChatGPT, Claude, Hugging Face, Jupyter +
Copilot
EDA and Feature Engineering with LLMs
Using LLMs to guide exploratory data analysis
Prompting for feature ideas and correlation detection
Automating data cleaning with prompt-driven code
Model Building and Code Generation
Generating code for regression, classification, or
clustering models
Prompting for hyperparameter suggestions and tuning
Explaining model choices and metrics using LLMs
MODEL EXPLANATION, SIMULATION, AND
AUTOMATION
Interpreting Model Results with LLMs
Explaining predictions and coefficients in plain language
Using LLMs to describe model accuracy, overfitting, and
bias
Summarizing performance across multiple models
Scenario Simulation and Predictive
Reasoning
Using LLMs to generate what-if scenarios
Prompting for domain-specific insights and assumptions
Limitations of LLMs in causal inference and uncertainty
Narrative Reporting and Presentation
Automation
Auto-generating executive summaries and visual insights
Integrating LLM output with BI tools, Notebooks, or slide
decks
Validating outputs and combining human + AI analysis
Final Labs and Wrap-Up
Discussion on production usage, ethics, and next learning
steps
