syn-data-csv 0.2.2

Synthetic data generation using LLMs

Synthetic Data Generation PoC

Overview

This project generates synthetic data using an AI model based on a given YAML configuration file and a reference CSV file. The generated data follows the provided column structure and user-defined constraints.

Features

• Reads column definitions and user instructions from a YAML configuration file.
• Validates the structure of the YAML file.
• Validates the format of the reference CSV file.
• Generates synthetic data using Groq's Mixtral model.
• Ensures the output is in CSV format without headers or extra text.
• Enforces unique rows with consistent column counts.

Prerequisites

• Python 3.8+
• Required Python packages:
  • yaml
  • pandas
  • json
  • csv
  • groq

Usage

Run the script with the required YAML configuration file and reference CSV file:

python test1.py config.yaml test_data.csv

Arguments

config.yaml - YAML file specifying column definitions and generation rules. test_data.csv - Reference CSV file to guide synthetic data generation.

YAML Configuration Structure

The YAML file should define columns and the prompt:

columns:
  - name: "id"
    type: "integer"
  - name: "name"
    type: "string"
  - name: "email"
    type: "string"
  - name: "age"
    type: "integer"
  - name: "city"
    type: "string"
  - name: "signup_date"
    type: "datetime"
    
num_rows: 100
prompt: "Generate a dataset for user profiles."

Reference CSV File (test_data.csv)

This file provides sample data for reference.

Example test_data.csv:

id,name,age,email,city,signup_date
101,John Doe,28,[e-mail],New York,2023-05-14
102,Jane Smith,34,[e-mail],Los Angeles,2022-11-21
...

Expected Output

• The generated dataset will be stored in <file_name>_synthetic.csv.
• The output will contain only comma-separated values without extra text or headers.
• The generated rows will be unique and follow the reference data pattern.

Error Handling

• If the YAML configuration is invalid, the script will display an error message and exit.
• If the reference CSV file has missing columns, the script will halt with an error.
• If no valid data is generated, an appropriate message will be displayed.

License

This project is for demonstration purposes only. Usage is subject to Groq API policies.