Large Language Models for Code Generation: The Practitioners' Perspective

Objective

This repository provides the replication package.xlsx for the study "Large Language Models for Code Generation: The Practitioners’ Perspective." The study aims to explore how practitioners perceive and use large language models for code generation. We collected survey data and performed:

Descriptive Analysis for closed-ended survey questions (Q1 to Q13) and Open Coding for open-ended survey responses (Q14 and Q15).

The replication package is provided above under the name "Replication_Package.xlsx".

1️⃣ Survey Data

The dataset consists of responses collected from software development practitioners regarding their experience with large language models for code generation. The data includes:

Closed-ended questions (Q1-Q13): Responses related to experience, role, industry, programming languages, tool usability, and performance evaluation of different models.

Open-ended questions (Q14-Q15): Participants' insights on challenges and suggestions for improvements in code generation models.

📂 Files:

Survey_Data - Contains the raw survey responses.

2️⃣ Descriptive Analysis

We performed descriptive statistical analysis on the closed-ended questions (Q1 to Q13) to extract insights such as:

Distribution of participants' experience and industry.

Preferences for programming languages used in testing and development.

Feedback on tool usability and performance of different models.

📂 Files:

Descriptive_Analysis - Contains summary statistics and key insights from Q1 to Q13.

3️⃣ Open Coding

For open-ended survey questions (Q14 and Q15), we applied thematic analysis and open coding to categorize qualitative responses into meaningful themes. This helped in identifying:

Common challenges faced by practitioners while using LLMs for code generation.

Suggestions for improving these models.

📂 Files:

OpenCoding - Contains coded responses and thematic analysis. A document explaining the coding process and framework used.

🔍 How to Use This Package

Researchers can use the survey data for further analysis or comparative studies.

The descriptive analysis provides insights into practitioners' experiences with LLMs.

The open coding results offer qualitative insights into user challenges and recommendations.

📩 Contact & Contributions

For questions or contributions, feel free to open an issue or submit a pull request. This dataset and analysis are shared for academic and research purposes. You can also contact on this email: [email protected]