Advanced Use Cases
Complex target function recipes that go beyond the basics. These recipes demonstrate advanced patterns such as multi-source joins, pandas integration, calendar arithmetic, rolling windows, and set-based exclusion logic.
Looking for simpler recipes?
See Core Recipes for introductory examples with full step-by-step explanations and both Python App / GUI App code styles.
Binary Classification
Predict a yes/no outcome per entity — with complex multi-source logic, cross-event joins, and calendar-aware conditions.
| Recipe | Industry | Description |
|---|---|---|
| User Silence Detection | Digital | Detect if a user goes silent across multiple event streams |
| IoT Sensor Offline | IoT | Predict if a sensor stays offline for 12+ continuous hours |
| Mobile Payment Adoption | Fintech | Identify new users who adopt mobile payments within 14 days |
| Biometric Login | Banking | Predict if a customer enables biometric login by month-end |
| Product Returns | E-commerce | Detect product returns within 30 days of delivery |
| Extended Warranty | Retail | Predict warranty purchase within 7 days of buying a laptop |
| Positive Reviews | E-commerce | Will the customer leave positive reviews for all items in next order? |
| App Channel Shift | Banking | Detect shift to 80%+ app-based transactions |
| Weekday Purchase | Retail | Predict online purchase on a specific weekday |
| Installment Defaults | Finance | Predict if customer misses >3 installment deadlines in 6 months |
| In-Game Purchase | Gaming | Predict in-game purchase within a new player's first 5 sessions |
| Subscription Churn | Fitness | Detect churn risk from reduced activity while subscription is active |
| Course Completion | EdTech | Predict course completion without premium subscription purchase |
Multiclass Classification
Predict which single class best describes the entity — using probability distributions and fiscal-period logic.
| Recipe | Industry | Description |
|---|---|---|
| Weekend Card Channel | Banking | Predict the dominant weekend transaction channel as softmax probabilities |
| Spending Tier | Retail | Classify customer into spending tiers based on quarterly history |
Multilabel Classification
Predict multiple independent outcomes per entity — with time-of-day filtering, rolling windows, and multi-channel detection.
| Recipe | Industry | Description |
|---|---|---|
| Evening Brand Purchases | Retail | Predict which brands a customer buys 2+ times during evening hours |
| Weekly Category Purchases | Retail | Predict which categories a customer buys from every week over 12 weeks |
| Ticket Escalation Channels | Support | Predict which channels a support ticket will escalate through |
Regression
Predict a continuous value per entity — time-to-event, counts, durations, and peak values.
| Recipe | Industry | Description |
|---|---|---|
| Days Until Complaint | General | Predict days until first customer complaint within 90 days |
| Buy Online Return In-Store | Retail | Count cross-channel buy-online-return-in-store events |
| New Category Purchase Time | Retail | Predict days until customer tries a new product category |
| Peak Daily Data Usage | Telecom | Predict highest daily mobile data usage in 30 days |
| Days with Long Calls | Telecom | Count days with calls exceeding 20 minutes |
| Training Duration | Fitness | Predict total training time excluding short sessions |
| New Lesson Duration | EdTech | Predict total duration of new (not repeated) lessons |
Recommendation
Produce a ranked list of items per entity — with exclusion logic, discount-based filtering, and category-aware selection.
| Recipe | Industry | Description |
|---|---|---|
| Top Repurchase | Retail | Recommend products likely to be repurchased, excluding recent buys |
| Never at Full Price | Retail | Recommend products previously only bought with discounts |
| Active Categories | Retail | Recommend unseen products from frequently purchased categories |