Machine Learning

This hub focuses on machine learning as an operational discipline: evaluation, monitoring, calibration, data quality and the failure modes that appear after a model leaves a notebook.

Start here

Monitoring and drift

Distribution shift, concept drift, model monitoring, base-rate drift and fairness under changing populations.

Data Drift and Fairness: Monitoring Equity When Populations Change

A fair model at launch can become unfair in production when populations, behavior, policies, or measurement systems change.

Using Unsupervised Learning for Early Data Drift Detection

When labels arrive late, production teams need early signals that the input environment has changed. Unsupervised drift detection can provide those signals, but only if it is designed around operations rather than dashboards.

Data Drift vs. Concept Drift: Understanding the Differences and Implications

Learn how data drift and concept drift can degrade machine learning models over time, and why continuous monitoring and adaptive systems are essential for model performance.

How to Detect Data Drift in Machine Learning Models

Data drift is one of the primary threats to model reliability in production. This article walks through how to detect it using both statistical techniques and modern monitoring tools.

Techniques for Monitoring and Managing Model Drift in Production

Model drift is inevitable in production ML systems. This guide explores monitoring strategies, alert systems, and retraining workflows to keep models accurate and robust over time.

Model Drift: Why Even the Best Machine Learning Models Fail Over Time

Even the best machine learning models experience performance degradation over time due to model drift. Learn about the causes of model drift and how it affects production systems.

Understanding Data Drift: What It Is and Why It Matters in Machine Learning

Data drift can significantly affect the performance of machine learning models over time. Learn about different types of drift and how they impact model predictions in dynamic environments.

Solving Data Drift Issues in Credit Risk Models

A comprehensive exploration of data drift in credit risk models, examining practical methods to identify and address drift using multivariate techniques.

Sequential Detection of Switches in Models with Changing Structures

Learn about sequential detection techniques for identifying switches in models with changing structures. Explore methods for detecting structural changes in time-series data and dynamic systems.

Managing Covariate Shifts in Machine Learning Models

Learn how to manage covariate shifts in machine learning models through effective model monitoring, feature engineering, and adaptation strategies to maintain model accuracy and performance.

Evaluation and uncertainty

Calibration, model evaluation, conformal prediction, abstention, slice-based evaluation and threshold decisions.

Global vs Local Models in Time Series Forecasting

The traditional approach fits one model per series. Modern practice often fits a single model across thousands of them, and usually wins.

Causal Feature Selection for Observational Machine Learning

Predictive feature selection is not enough when a model supports interventions. This article explains how causal thinking improves feature design in observational machine learning.

Probabilistic Forecasting: Beyond the Point Estimate

A point forecast answers the wrong question. Most decisions depend on how bad things could plausibly get, which is a statement about the whole distribution.

Forecasting Baselines That Are Hard to Beat

An RMSE of 4.2 means nothing on its own. Without a baseline you cannot tell whether a model is skilful or merely arithmetic.

Cost-Sensitive Learning for Rare Event Prediction

Rare event models should be optimized for decisions, not only class balance. Cost-sensitive learning connects model thresholds to real operational consequences.

Decision Curve Analysis: Measuring Whether Predictive Models Are Worth Acting On

Decision curve analysis evaluates predictive models by asking whether acting on their predictions produces better decisions than simple alternatives.

Label Noise in Supervised Learning: When the Target Cannot Be Trusted

Label noise is one of the most damaging data quality problems in supervised learning because it corrupts the target the model is trained to imitate.

Slice-Based Model Evaluation: Finding the Failures Average Metrics Hide

Slice-based evaluation exposes where a machine learning model fails by breaking aggregate performance into meaningful subgroups, conditions, and operational contexts.

Temporal Validation in Machine Learning: Testing Models Against the Future

Temporal validation evaluates machine learning models the way they will be used: trained on the past and tested on the future.

Probability Calibration in Machine Learning: From Classical Methods to Modern Approaches and Venn–ABERS Predictors

Explore the evolution of probability calibration methods in machine learning, from histogram binning to Venn–ABERS predictors, with a deep dive into theory, implementation, and applications.

Feature and data quality

Feature engineering, leakage, missing data, weak supervision, label noise and cost-sensitive learning.

Feature Engineering for Time Series Without Leaking the Future

Turning a time series into a tabular problem unlocks powerful models and introduces a specific failure: features that quietly contain information from the future.

Missing Data Mechanisms in Machine Learning

Missing data is not only a preprocessing nuisance. The reason data is missing can change model bias, fairness, monitoring, and deployment behavior.

Fourier Analysis for Data Science: From Signals to Features

Fourier analysis is more than a signal-processing trick. It is a way to ask which cycles, rhythms, and scales explain variation in data.

Representation Learning for Tabular Data: Beyond Manual Feature Engineering

Representation learning for tabular data is not about replacing feature engineering blindly. It is about learning useful structure while respecting the constraints of business data.

Traffic Prediction: Advanced Analytics for Smart Transportation Systems

A comprehensive guide to traffic prediction in smart transportation systems, covering data sources, preprocessing, modeling approaches, and real-world Python examples.

Crafting Time Series Features for Better Models

Learn specialized feature engineering techniques to make time series data more predictive for machine learning models.

Least Angle Regression: A Gentle Dive into LARS

Least Angle Regression, or LARS, is an efficient regression algorithm designed for high-dimensional data. It provides a pathwise approach to linear regression that is especially useful in the presence of multicollinearity or when feature selection is crucial.

Handling Non-Stationarity in Time Series Data: Techniques and Best Practices

Non-stationarity is one of the biggest challenges in time series analysis. Explore proven techniques and statistical tools to transform non-stationary data into model-ready series.