AI-Driven Auction Pricing System Development

I need to built an AI-driven system that dynamically estimates optimal auctions starting prices to addresses the inefficiencies in the current manual and static reserve-pricing system for vehicle plate auctions on the Absher platform. It is divided into two distinct projects that work in tandem: the AI Engine and the Analytics Dashboard

Project 1: The AI Pricing Engine (Backend)

This is the "Brain" of the operation. It handles the heavy lifting of mathematical modeling and real-time forecasting.

Primary Purpose: To transform cultural plate features into monetary value using Machine Learning and to forecast auction trajectories.

Core Capabilities:

Dynamic CART Regression: Uses the Decision Tree algorithm to calculate reserve prices based on non-linear features (Symmetry, Repetition).

Bid Velocity Forecasting: A time-series logic engine that predicts the final closing price based on the speed and intensity of incoming bids.

Live Model Re-training (Online Learning): Allows the model to be updated via API at runtime without a server restart.

Project 2: The Analytics Dashboard (Frontend)

This is the "Control Tower." It provides a user-friendly interface for administrators to interact with the AI and visualize data.

Primary Purpose: To provide a visual simulation environment for "What-if" analysis of Saudi vehicle plate auctions.

Core Capabilities:

Real-time Visualization: Uses interactive charts to show the relationship between the starting reserve and the predicted final outcome.

Simulation Controls: inputs to adjust plate variables (Seasonality, Symmetry) at runtime.

AI Lab (Admin Panel): A management table that allows the user to see, modify, and "inject" market truth directly into the AI's memory.

Processing: The backend runs the CART algorithm. If data was changed, it re-trains the model in milliseconds.

Forecasting: The backend calculates the "Bid Velocity" based on current auction parameters.