“It doesn’t guess. It learns — and then it loops.”
A deterministic, self-refining insight engine that transforms raw operational data into structured, actionable intelligence.
No cloud dependencies. No black-box AI. Just Python, Flask, and clarity.
The RPLE ingests structured metadata (CSV), identifies emerging patterns, filters noise, and outputs continuously improving predictions and insights.
Each cycle compounds analytical clarity — meaning the more you use it, the smarter it gets.
Built for transparency, not mystery: every decision is logged, scored, and explainable.
- 📊 CSV Ingestion & Normalization — feed any structured operational dataset
- 🔁 Recursive Insight Loop — learns from every cycle, compounding precision
- 🧩 Constructive Feedback Engine — outputs clear insights, not just numbers
- 🧠 Memory Reservoir — tracks persisting vs. novel patterns over time
- 🪶 Lightweight Deployment — pure Python + Flask, no external APIs or services
graph TD
A[CSV Upload] --> B[Data Normalization]
B --> C[Pattern Detection & Scoring]
C --> D[Insight Generation]
D --> E[Human Feedback + Memory Reservoir]
E --> C
-
Upload a CSV file (any dataset with
value_1,value_2, and optionalriskcolumns). -
The engine analyzes:
- Trends
- Correlation shifts
- Anomalies
- Risk alignments
-
It generates insight cards with:
- Confidence
- Novelty
- Severity
- Suggested actions
- Status (🆕 new / ♻️ persisting)
-
The engine stores each insight in its memory reservoir for future comparison.
| Domain | Insight | Confidence | Novelty | Status |
|---|---|---|---|---|
| Primary | Primary metric trending up | 0.86 | 0.82 | 🆕 |
| Correlation | Relationship between value_1 and value_2 strengthened | 0.72 | 0.75 | ♻️ |
| Anomaly | Anomaly burst detected (3 spikes) | 0.91 | 0.88 | 🆕 |
| Risk | Risk and primary metric are aligned (corr=0.52) | 0.67 | 0.64 | 🆕 |
git clone https://github.com/<your-handle>/recursive-logic-engine.git
cd recursive-logic-engine
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python app.pyThen open your browser at http://127.0.0.1:5000 and upload your dataset.
sample_input.csv
timestamp,value_1,value_2,risk
2025-10-01,10,12,0.20
2025-10-02,11,12,0.21
2025-10-03,13,13,0.22
2025-10-04,15,14,0.25
2025-10-05,18,14,0.30
2025-10-06,16,13,0.28
2025-10-07,19,15,0.33
2025-10-08,21,16,0.36
2025-10-09,24,17,0.40
Each loop stores a hashed summary of every insight in insight_memory.json.
Future runs detect whether insights are:
- 🆕 New: unseen patterns
- ♻️ Persisting: confirmed patterns continuing across cycles
This creates a real-time feedback model that grows smarter with use.
- Python 3.9+
- Flask 3.x
- Pandas + NumPy
- JSON + Local Storage (no external API)
- SHA-256 integrity for insight memory
Joseph Wells 📍 Indianapolis, IN 📧 joepwells95@gmail.com 🔗 Foxhunter Labs
- 🦊 Foxhunter Pro — Human-Gated Reconnaissance & Ethical Autonomy System
- 🧬 Enigma² — Safety & Kill-Switch Engine
- 🛰️ Swarm — Deterministic Multi-Agent Coordination Framework
MIT License © 2025 Joseph Wells Use freely for educational and research purposes. Attribution required.
“Predictive clarity doesn’t just happen — it compounds.”