🚀 PDP – Profile Dataset Processor

AI-powered performance dataset analyzer for Unreal Engine profiling workflows.

---

📌 Overview

PDP (Profile Dataset Processor) is a tool designed to analyze Unreal Engine performance datasets (CSV profiling data) using locally hosted Large Language Models (LLMs).

It converts raw profiling data into structured insights, helping developers quickly identify performance bottlenecks, understand root causes, and apply fixes efficiently.

---

✨ Features

• 📊 CSV Dataset Analysis

  • Supports Unreal Engine profiling exports
  • Handles large datasets efficiently

• 🧠 AI-Powered Insights

  • Uses locally hosted LLMs (privacy-friendly)
  • Generates detailed analysis reports

• 📈 Statistics Generation

  • Frame time analysis
  • CPU/GPU breakdown
  • Bottleneck detection

• ⚠️ Issue Detection

  • Detects spikes and anomalies
  • Highlights performance issues

• 🔍 Root Cause Analysis

  • Explains why issues occur
  • Connects metrics to real-world problems

• 🛠️ Suggested Fixes

  • Actionable optimization tips
  • Unreal Engine–focused recommendations

• 🎨 Structured HTML Reports

  • Organized output
  • Easy-to-read analysis sections

---

🧩 Unreal Engine Plugin

PDPC includes an Unreal Editor Toolbar Plugin.

🔘 Features inside Unreal:

• Start Capture

  • Begins CSV profiling capture

• Stop & Launch PDPC

  • Stops capture
  • Automatically launches PDPClient

---

🛠️ Tech Stack

• Python
• Pandas
• Matplotlib
• Tkinter
• Local LLM API (Ollama)

---

⚙️ Installation

1. Clone the Repository

git clone https://github.com/sayoojjs/Profile_Data_Processor.git

2. PDP Plugin Setup

Copy the plugin folder into:

YourProject/Plugins/

Open Unreal Engine

Go to Edit → Plugins
Enable PDPClient

Restart the editor

3. Python Setup

Install required dependencies:

pip install pandas matplotlib tkhtmlview requests

4. Setup Local LLM

• Ollama

Make sure your local model server is running.

▶️ Usage

From Unreal Engine:

• Go to the menubar and click Profile Data Processor
• Click Start Capture
• Run your scene/game
• Click Stop & Launch PDPC

From PDPClient:

• Set server address
• Select the language model that you prefer to use
• Run analysis
• View generated HTML report
• Export as HTML for better viewing