Predictive Analytics
Forecasting and predictions
Recommendation System Design Prompt
Design a recommendation system that solves the cold start problem, handles sparse interaction data, and optimizes for the business metric that matters — not just click-through rate — with an offline evaluation framework that predicts online performance.
NLP Text Classification Model Prompt
Build a text classification pipeline that selects the model architecture based on dataset size and label structure — from TF-IDF baselines to fine-tuned transformers — with a calibration strategy that makes confidence scores meaningful.
Anomaly Detection System Design Prompt
Design an anomaly detection system that distinguishes genuine anomalies from noise — with method selection by anomaly type, threshold calibration to the false positive budget, and an alert triage framework that prevents alert fatigue.
Clustering & Customer Segmentation Prompt
Design a clustering and segmentation analysis that produces actionable, stable segments — with algorithm selection by data structure, optimal cluster count determination, and a segment profiling framework that translates statistical clusters into business-meaningful groups.
Model Evaluation & Validation Framework Prompt
Design a model evaluation framework that measures genuine out-of-sample generalization — with validation strategy selection, metric choice tied to the business objective, and a diagnostic protocol that distinguishes overfitting from underfitting from distribution shift.
Feature Engineering Framework Prompt
Design a feature engineering pipeline that creates predictive signal from raw data — with domain-driven feature hypotheses, transformation specifications, leakage prevention, and a selection protocol that keeps only features that improve out-of-sample performance.
Customer Churn Prediction Model Prompt
Build a churn prediction model that identifies at-risk customers early enough to intervene — with feature engineering from behavioral signals, survival analysis for time-to-churn, and a business-impact framework that prioritizes interventions by expected retention value.
Time Series Forecasting Model Prompt
Design a time series forecasting pipeline that selects the model based on the series' actual temporal structure — trend, seasonality, stationarity, and external drivers — not the model that's currently popular.
Classification Model Framework Prompt
Build a classification model framework that selects the right algorithm, calibrates the decision threshold to the business cost structure, and evaluates performance on metrics that match what the model is actually used for.
Regression Model Selection & Specification Prompt
Select and specify the right regression model for your prediction problem — with assumption checks, regularization decisions, and a validation framework that distinguishes genuine predictive power from overfitting.