koushik-marka/campaigniq

ML platform for bank marketing campaign optimisation — real-time customer scoring, batch CSV processing, ROI simulation · ROC-AUC 0.800 · Deployed on GitHub Pages # CampaignIQ — Bank Marketing Intelligence Platform > A production-ready ML dashboard for predicting bank term deposit subscriptions and optimising campaign ROI. **Built by M. Koushik** · SR University · Dataset: UCI Bank Marketing (41,188 customers) --- ## Live Demo 🔗 koushik-luffy.github.io Open the link and start using the platform instantly — no setup required. --- ## What This Does CampaignIQ turns a trained ML pipeline (Logistic Regression, Random Forest, XGBoost) into a fully interactive web platform that bank campaign managers can use **without writing a single line of code**. | Feature | Description | |---|---| | **Real-time Predict** | Score any individual customer instantly — change any field and probability updates live | | **Batch Scoring** | Upload a CSV of your contact list → every row gets a probability score + segment label → export back | | **Targeting Segments** | Auto-segments customers into High / Medium / Low priority with recommended call actions | | **ROI Calculator** | Sliders for cost-per-call and revenue-per-subscription — compares random calling vs ML-targeted net profit | | **Model Lab** | Side-by-side metrics, ROC curves, confusion matrix inspector for all 3 models | | **Business Insights** | EDA findings, 5-step campaign workflow, dataset expansion roadmap | --- ## ML Results | Model | Precision | Recall | F1 | ROC-AUC | |---|---|---|---|---| | Logistic Regression | 0.365 | 0.643 | 0.465 | **0.800** ✅ Best | | Random Forest | **0.552** | 0.275 | 0.367 | 0.780 | | XGBoost | 0.373 | 0.585 | 0.455 | 0.775 | **Key business result:** Targeting only the high-priority segment (17% of customers) achieves a **41.3% subscription rate** vs 11.3% overall — a **3.7× efficiency gain**. --- ## Project Structure ``` campaigniq/ ├── index.html ← Entire application (self-contained) ├── README.md ← This file ├── LICENSE ← MIT License ├── .gitignore ← Standard gitignore ├── docs/ │ …