AI Ad Performance Optimizer

PyQt5: for Ad Analytics, Optimization, and Insights

---

Overview

The AI Ad Performance Optimizer is a comprehensive desktop application built with Python, PyQt5, scikit-learn, Stable-Baselines3 (RL), and Matplotlib. It helps digital marketers and advertisers analyze ad campaigns, predict CTR & conversions, optimize ad budgets, and generate actionable insights—all from a single CSV dataset.

This tool integrates ML prediction, RL-based budget optimization, NLP sentiment analysis, and interactive visualizations to make ad performance management smarter and faster.

---

Key Features

Data Handling

• Load ad campaign data from CSV files.
• Automatically normalize and clean columns (Ad_ID, Budget, CTR, Conversions, etc.).
• Handle numeric coercion for financial and performance metrics.

Machine Learning

• Predict CTR (Click Through Rate) and Conversions for ads using ML fallback.
• Generate automated insights highlighting low-performing ads.

Explainable AI (XAI)

• Compute feature importance using Random Forest models.
• Identify which ad metrics (e.g., Budget, Impressions, Clicks) influence performance most.

Reinforcement Learning (RL)

• Optimize ad budgets across campaigns using PPO (Proximal Policy Optimization).
• Users can select the metric to maximize: CTR, Conversions, or minimize Cost.
• View optimized budget allocation in interactive plots.

NLP Ad Analysis

• Analyze ad text sentiment (Positive, Neutral, Negative).
• Provide ad copy suggestions to improve engagement.

Scenario Simulation

• Simulate "what-if" scenarios like budget increases.
• Evaluate impact on predicted CTR and conversions.

Interactive Dashboard

• Visualize metrics like CTR per Product, Conversions per Region, Budget allocation.
• Includes zoom/pan toolbar and save plot functionality.

Export Reports

• Export processed data and insights to CSV or Excel.
• Includes predicted metrics, optimized budget, sentiment analysis, and suggestions.

---

Installation

1. Clone the repository:

git clone https://github.com/Mohima6/AI-Ad-Performance-Optimizer.git
cd AI-Ad-Performance-Optimizer

2. Install required packages:

pip install -r requirements.txt

Requirements: Python ≥3.10, PyQt5, pandas, numpy, matplotlib, scikit-learn, stable-baselines3, gym, shimmy

3. Run the application:

python app/main.py

---

Usage

1. Load CSV: Click "Load CSV" and select your ad campaign dataset.
2. Predict Performance: Click "Predict CTR/Conversions" to generate predictions.
3. Optimize Budget: Click "Optimize Budget (RL)" and choose a metric to maximize.
4. Text Analysis: Click "Analyze Ad Text" for sentiment and copy suggestions.
5. Scenario Simulation: Click "Scenario Simulation" to test budget changes.
6. XAI Feature Importance: Click "XAI Feature Importance" to see which features matter most.
7. Export Report: Click "Export Report" to save your processed dataset.

Credits

• Python, PyQt5, pandas, numpy, matplotlib, scikit-learn, stable-baselines3, gym
• Inspired by the need for smarter ad budget optimization and campaign insights

---

Problem Statement & Motivation

• Running effective digital advertising campaigns is challenging because:

• Marketers often have hundreds of ads across multiple platforms, making manual analysis time-consuming.

• Budgets are limited, and inefficient allocation can drastically reduce ROI.

• Identifying high-performing ads and understanding what influences performance is difficult without proper analytics.

• Ad copy and messaging often lack data-driven improvements, which can hurt engagement.

This project solves these problems by-

• Automatically predicting CTR and conversions for all ads in a dataset.

• Providing RL-based budget optimization to maximize returns from limited resources.

• Offering explainable insights (XAI) on what metrics drive ad performance.

• Suggesting improvements for ad copy based on sentiment analysis.

• Allowing interactive visualization and scenario simulations for better decision-making.

In short: It helps marketers save time, improve ROI, and make data-driven decisions with ease.

---