DM VEE Exception Auto-Resolver Agent

An intelligent agent for automatically detecting and resolving exceptions in interval meter data (MDM - Meter Data Management).

What This Project Does

The MDM VEE (Validation, Estimation, and Editing) Exception Auto-Resolver is a Python-based system that:

1. Generates realistic synthetic interval meter data with daily patterns, temperature sensitivity, and weekend effects
2. Detects common exceptions including missing intervals, spikes, and register rollovers
3. Proposes intelligent fixes using rule-based heuristics, SARIMAX time series forecasting, and neighbor meter profiling
4. Applies fixes automatically with confidence scoring and constraint validation
5. Maintains a complete audit trail of all applied fixes with explanations
6. Evaluates performance using standard metrics (MAE, MAPE, RMSE)
7. Provides a web dashboard for visualization and analysis

Quick Start

Prerequisites

• Python 3.10+
• pip

Installation

# Clone the repository
git clone <repository-url>
cd mdm-vee-agent

# Create virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

Basic Usage

# 1. Generate synthetic data
python scripts/generate_synthetic_data.py --meters 100 --days 30 --out data/raw/clean.parquet

# 2. Inject exceptions
python scripts/inject_exceptions.py --in data/raw/clean.parquet --out data/interim/with_exceptions.parquet --exceptions data/interim/exceptions.jsonl

# 3. Run the complete agent pipeline
python scripts/run_agent.py --input data/interim/with_exceptions.parquet --output data/processed/fixed.parquet --audit data/processed/audit_log.jsonl --confidence-threshold 0.9

# 4. Evaluate performance
python scripts/evaluate_agent.py --fixed data/processed/fixed.parquet --truth data/raw/clean.parquet

# 5. Launch the dashboard
streamlit run scripts/demo_streamlit_app.py