My tags: topic: terminology#
With this tag
- Subsampling
- Class Weighting
- SMOTE (Synthetic Minority Over-sampling Technique)
- Oversampling
- Low-pass Filtering
- NearMiss (Distance-based Undersampling)
- Cluster-based undersampling
- Random Undersampling
- Signal Processing
- Time Series
- Micro AUROC
- Multi-label Classification
- Micro F1
- Single-label Classification
- Micro Recall
- Micro Precision
- One-vs-Rest (OvR) AUROC
- Macro AUROC (Macro-Averaged AUROC)
- Macro F1
- Macro Recall
- Macro Precision
- Multiclass AUROC
- Gini Coefficient
- Bootstrap Confidence Intervals (CIs)
- Probability
- Mann–Whitney U Test (also called the Wilcoxon rank-sum test)
- Predictive Parity (Calibration)
- Equalized Odds (Fairness)
- Equal Opportunity (Fairness)
- Demographic Parity (Statistical Parity)
- Cross-Selling
- Upselling
- Customer Segmentation
- SaaS (Software as a Service)
- Valuation Metric
- D2C (Direct-to-Consumer)
- LTV:CAC Ratio
- Net LTV (sometimes called Contribution LTV)
- Gross LTV (Customer Lifetime Value)
- Predictive LTV (pLTV)
- Cohort-Based LTV (Simple Version)
- Customer Lifetime
- Gross Margin
- Fully Loaded CAC (Customer Acquisition Cost)
- Organic CAC (Customer Acquisition Cost)
- Paid CAC (Customer Acquisition Cost)
- Channel-Specific CAC (Customer Acquisition Cost)
- Blended CAC (Customer Acquisition Cost)
- Lead-Gen Software
- Thompson Sampling (TS) in Bandits (Multi-Armed Bandit Problem (MAB))
- Bayesian Decision Theory (BDT)
- Bayesian Time Series
- Posterior probability of uplift
- Gaussian Processes (GPs)
- Bayesian Neural Networks (BNNs)
- Variational Inference (VI)
- MCMC (Markov Chain Monte Carlo)
- Sequential Settings
- Frequentist
- Binomial Likelihood
- Posterior belief
- Marginal Likelihood (also called The Model Evidence or Integrated Likelihood)
- Posterior
- Prior Belief (or Prior Probability)
- Parameter(s) of Interest
- Bayes’ Theorem
- Conversion Rate Uplift
- Bayesian Stopping Rules
- Optimizely
- Online Experimentation Platforms
- Stopping Rules
- Treatment Effect
- Posterior Probability
- Bayesian Sequential Testing
- Likelihood Ratio (LR)
- Sequential Probability Ratio Test (SPRT)
- Pocock Method
- O’Brien–Fleming (OBF) Method
- Group Sequential Testing
- Type I Error
- Traditional A/B Test (Fixed-Horizon A/B Test)
- Fixed-Horizon Testing
- True Conversion Rate
- Standard Error (SE)
- True Mean (Population Mean)
- Margin of Error (MoE)
- Critical Value
- Sample Standard Deviation
- Sample Mean
- Regression Coefficient
- Proportion
- True Population Parameter
- Compromise Power Analysis
- Post Hoc Power Analysis
- A Priori Power Analysis
- Statistical Significance
- Z-Score
- Two-Proportion Z-Test
- Beta Distribution
- Google Experiments
- Minimum Detectable Lift (MDL)
- Trivial Effects
- Sample size
- Power (1 – β)
- Significance Level (α)
- Effect Size (δ)
- Hypothesis Testing
- Ranking Algorithms
- Probabilistic Interleaving
- Team Draft Interleaving (TDI)
- Balanced Interleaving
- Causal Impact
- Bandit Algorithms
- A/B/n Test
- Multivariate Test (MVT)
- Risk of Peeking
- Causal Inference
- P-Value (probability value)
- Z-Test
- T-Test
- Session Length
- Revenue per User (RPU / ARPU)
- Churn
- Retention
- Statistically Significant
- IID (Independent and Identically Distributed)
- Temporal autocorrelation (Serial Correlation)
- Blocked Splits (Single Holdout)
- Sliding Window (Rolling Window) Cross-Validation
- Expanding Window Cross-Validation
- Data Leakage
- Stratified Group K-Fold
- Stratified Shuffle Split
- Multiclass stratified CV
- k-fold cross-validation
- Cross-Validation (CV)
- Re-scoring
- Drift Detection
- Model Distillation (Knowledge Distillation)
- Early Stopping
- Epochs
- Hyperparameter
- AI (Artificial Intelligence)
- Machine Learning (ML)
- Medical AI
- KYC
- FTEs
- AWS SageMaker
- Vertex AI
- OpenAI API (ML API)
- AWS SageMaker Endpoints
- Cloud Inference with Big Payloads
- Cloud Inference
- Ensemble
- Model Weights
- FLOPs
- OpEx
- LLMs (Large Language Models)
- Recalibration
- Reweighting
- Continuous Retraining
- Monitoring Pipelines
- Active Learning
- Bayesian Correction
- Recalibrate Thresholds
- Guardrails (in ML & Data Systems)
- Model KPIs (Key Performance Indicators)
- Lagging Indicators
- Leading Indicators
- Windows (in Time-Series)
- Autoencoder
- Frozen Encoder
- Embedding
- Representation Shift
- Classifier Two-Sample Tests (C2STs)
- Energy Distance
- Maximum Mean Discrepancy (MMD)
- Cardinality in Categorical Data
- Categorical Drift
- Cramér’s V
- Macro Shifts
- Categorical Explosions
- Cohort
- Off-Distribution
- Discriminatory Power
- KS Statistic (Kolmogorov–Smirnov Statistic)
- Model Stability
- Feature Values
- Four-Fifths (80%) Rule
- SLI (Service Level Indicator)
- ROI (Return on Investment)
- Treatment Cost
- Incremental Revenue
- Incremental Recovery Rate (IRR)
- Incremental Sales
- Random Targeting Strategy
- Causal ML (Causal Machine Learning)
- Cumulative Uplift
- Population Proportion
- Incremental Gain
- Total Incremental Benefit (TIB)
- Cumulative Incremental Gain (CIG)
- Qini Curve
- Uplift Score
- Uplift Models
- Ops Health Dashboard
- SLA Breach Rate
- SLA (Service Level Agreement)
- Supplier Constraints
- Long Lead Times
- Slow-Moving SKUs
- SKU
- Real-Time Inventory Tracking
- Supplier Management
- Demand Forecasting
- Reorder Point (ROP) Optimization
- Safety Stock
- Backorder Rate
- Lost Sales Value
- Fill Rate
- Stockout Rate
- Prophet — Time Series Forecasting by Facebook (Meta)
- LSTM — Long Short-Term Memory Networks
- ARIMA (AutoRegressive Integrated Moving Average)
- Return Distribution
- Value-at-Risk (VaR)
- Risk Forecast
- Probabilistic Scoring
- Full Distribution
- Continuous Probabilistic Forecasts
- Classification Probability
- Quantile Forecasts
- Point Forecasts
- Strictly Proper Scoring Rules
- Probability Forecasts
- Target Variable
- Probability Density
- Normal Distribution
- Probability Mass
- Probability Distribution
- Probabilistic Forecasts
- Deterministic forecasts
- Cumulative Distribution Function (CDF)
- M-Competitions (Makridakis Competitions)
- Forecasting Benchmarks
- Average Absolute Error (AAE)
- Seasonal Lag
- Simple Baseline Methods
- Naïve Baseline Forecast
- Forecast Error
- Forecasting Competitions
- Predicting Percentiles
- Prediction Intervals (PI)
- Quantile Regression
- Quantile Level
- Time Series Forecasting
- Log-Space
- Relative accuracy
- R² (R-squared)
- Long-Tail Items
- Self-Information of Popularity
- Relevance in Recommender Systems
- Genre Overlap
- Jaccard index
- Cosine Similarity of Item Features
- Intra-List Diversity (ILD)
- Dominating in Recommender Systems
- Catalog Coverage
- User Coverage
- Item Coverage
- Diminishing Utility
- DCG (Discounted Cumulative Gain)
- Kaggle
- TREC (Text REtrieval Conference)
- Adaptive ECE (Expected Calibration Error with Adaptive Binning)
- Maximum Calibration Error (MCE)
- ROC Curve (Receiver Operating Characteristic)
- Murphy’s Decomposition
- Temperature Scaling
- Platt Scaling
- Isotonic Regression
- Support Vector Machines (SVMs)
- Underconfident
- Overconfident
- Confidence Level
- Risk-Based Decisions
- Neural Networks
- Binary Cross-Entropy (BCE)
- Loss Functions
- Underflow
- Logit Space
- Logistic Regression
- Binary Classification
- Classification Models
- Log-Odds
- Softmax Function
- Sigmoid Function
- Squashing Function
- Conversion Rate (CR)
- Cost-Per-Click (CPC) Models
- Causal Trees
- Uplift Random Forests
- Uplift Curve
- Likelihood
- Correlation
- Causal Effect
- Outlier
- Mean Squared Error (MSE)
- Regression Models
- One-vs-Rest (OvR)
- Multiclass Classification
- Partial AUC (pAUC)
- Micro AUC
- Macro AUC
- Median
- Mean
- Sensitivity in Feature Engineering
- Encode (in Feature Engineering)
- Normalize (in Feature Engineering)
- Embedding Similarity
- Computer Vision (CV)
- Natural Language Processing (NLP)
- Accuracy
- Chi-square (χ²) Test
- Kolmogorov–Smirnov (KS) Test
- Jensen–Shannon (JS) Divergence
- Kullback–Leibler (KL) Divergence
- Statistical Tests
- Seasonality
- Concept Drift
- Data Drift
- Fair Lending laws
- Basel III
- High-Stakes Domains
- Deep Ensembles
- Counterfactual Explanations
- LIME (Local Interpretable Model-agnostic Explanations)
- SHAP (SHapley Additive exPlanations)
- Post-hoc Explainability
- Decision Trees
- Linear Models
- Caching
- Quantization
- ONNX (Open Neural Network Exchange)
- Full Annotation
- Weak Supervision
- TPU Clusters
- Statistical Power
- Drift Guardrails
- Latency Guardrails
- Fairness Guardrails
- DeLong’s Test
- Dataset Shift
- Label Noise
- Evaluation Set
- Clopper–Pearson Interval
- Wilson Score Interval
- Per-class Precision (sometimes called class-wise precision)
- Multiclass Precision
- Multilabel Precision
- Weighted Averaging
- Harmonic Mean
- F1-score
- Model Score
- Bootstrap
- Average Precision (AP)
- Upsampling
- Downsampling
- Micro Averaging
- Macro Averaging
- AUC (Area Under the Curve)
- Fairness parity
- LTV (Customer Lifetime Value)
- CAC (Customer Acquisition Cost)
- Bayesian Inference.
- Sequential Testing (also called sequential analysis)
- Confidence Intervals (CIs)
- Power Analysis
- Interleaving Tests
- A/B Testing
- Time-based splits (a.k.a. Temporal Cross-Validation, Rolling Window Validation)
- k-fold Stratified Cross-Validation (Stratified CV)
- Compute budgets
- Manual review minutes
- Inference Cost (Inference $)
- Label Drift (a.k.a. Target Drift)
- Covariate Drift (a.k.a. Covariate Shift)
- KS shift (Kolmogorov–Smirnov shift)
- PSI (Population Stability Index)
- Selection Rate
- SLOs (Service Level Objectives)
- Cannibalization
- Revenue net of treatment cost
- Incremental Conversions
- Uplift@k
- AUUC (Area Under the Uplift Curve)
- Qini Coefficient
- Crew Overtime
- SLA Breaches
- Overstock %
- Stockouts
- Continuous Ranked Probability Score (CRPS)
- MASE (Mean Absolute Scaled Error)
- Pinball Loss (a.k.a. Quantile Loss)
- WMAPE (Weighted Mean Absolute Percentage Error)
- sMAPE (Symmetric Mean Absolute Percentage Error)
- RMSLE (Root Mean Squared Logarithmic Error)
- Mean Absolute Error (MAE)
- Novelty (in Recommender Systems)
- Diversity (in Recommender Systems)
- Coverage
- Hit Rate (HR)
- NDCG (Normalized Discounted Cumulative Gain)
- Mean Average Precision (MAP)
- Expected Calibration Error (ECE)
- Reliability Curves (also called Calibration Curves)
- Log Loss (also called Logarithmic Loss or Cross-Entropy Loss)
- Brier Score
- Calibration quality (Model Calibration)
- Logits
- CTR (Click-Through Rate)
- WAPE (Weighted Absolute Percentage Error)
- Recall
- Uplift
- Mean Absolute Percentage Error (MAPE)
- Root Mean Squared Error (RMSE)
- ROC-AUC (Receiver Operating Characteristic – Area Under Curve, = AUROC)
- Baseline Heuristics
- Precision (a.k.a. Positive Predictive Value, PPV)
- Precision–Recall AUC (PR-AUC)
- Advanced Sorting in Spreadsheets
- Terminology