Automating Data Manipulation Tasks in MS Excel: A Comparative Analysis of Gemini and Llama Language Models

This research examines various language models for generating python code that automates complex data manipulation tasks in Excel files, assessing models based on accuracy, speed, and accessibility. Managing large datasets and performing complex transformations in Excel can be challenging, hence Python-based automation simplifies these processes, by reducing time to write codes and its ability to manage huge datasets. To identify the most effective language models for generating code to automate such tasks, the models were compared from the Gemini and Llama families, using ROUGE scores to evaluate output quality. By prioritizing efficiency through speed and selecting lightweight models, our study bridges the research gap in choosing LLMs for automating Excel workflows, enhancing computational efficiency, supporting data analysis, and reducing time spent on repetitive tasks. Limitations included restricted access to language models leading to limited experimental scope.