In this episode, Lucas and Luna dive into gradient boosting, the machine learning technique behind countless winning models on Kaggle and in production systems. They explore how it works, why it's so effective, and the practical trade-offs teams face when using it. Using a concrete example of a churn prediction model at a telecom company, they break down the key hyperparameters like learning rate and tree depth, and explain how to avoid the classic pitfall of overfitting. They also discuss the differences between popular implementations like XGBoost and LightGBM, and when it might be worth reaching for a simpler model instead. Along the way, they touch on interpretability, feature importance, and the role of regularization. Whether you're a practitioner or just curious about what's under the hood of modern data science, this episode gives you a clear, grounded look at one of the most powerful tools in the field.

#GradientBoosting #XGBoost #LightGBM #MachineLearning #DataScience #ChurnPrediction #HyperparameterTuning #Overfitting #FeatureImportance #Interpretability #Kaggle #EnsembleMethods #TechPodcast #BusinessAnalytics #DataDriven #FexingoBusiness #BusinessPodcast #DataSciencePodcast

Keep every episode free: buymeacoffee.com/fexingo

In this milestone 150th episode, Lucas and Luna dive into probability calibration — the practice that makes a model's predicted confidence match real-world outcomes. They anchor the discussion with a concrete example: a fraud model that says '88 percent suspicious' but is right only 65 percent of the time. They explain why raw model outputs can be overconfident, especially in deep learning, and walk through practical fixes like Platt scaling and isotonic regression. Along the way, they touch on how calibration matters in high-stakes domains like credit risk and medical diagnosis, and why a well-calibrated model is often more trustworthy than a sharp but miscalibrated one. The episode is packed with actionable insights for data scientists and curious listeners alike, and it includes a brief, natural moment where the hosts reflect on keeping the podcast ad-free through listener support at buy me a coffee dot com slash fexingo.

#ProbabilityCalibration #ModelConfidence #PlattScaling #IsotonicRegression #MachineLearning #DataScience #FraudDetection #CreditRisk #MedicalDiagnosis #DeepLearning #Overconfidence #ModelReliability #Technology #DataSciencePodcast #FexingoBusiness #BusinessPodcast #Analytics #AI

Keep every episode free: buymeacoffee.com/fexingo

When a model's predictions start drifting from reality, the first question every data scientist asks is: where did the data go wrong? In this episode, Lucas and Luna explore a subtle but powerful debugging technique: data lineage. They walk through a real example at a mid-sized e-commerce company where a sudden drop in recommendation click-through rates turned out to trace back to a silent upstream schema change in a product catalog feed. Lucas explains how lineage graphs — the network of data sources, transformations, and downstream consumers — let teams trace a model's output all the way back to the raw inputs, and why most debugging setups only track half the story. The conversation also covers how lineage differs from data provenance, why it's often buried in orchestration logs and metadata catalogs, and how a simple lineage query can cut a multi-day investigation down to minutes. If you've ever stared at a stale model output and wondered what changed, this episode gives you a practical mental model for finding the answer.

#DataLineage #DataDebugging #DataPipelines #MachineLearning #DataScience #DataGovernance #Metadata #DataProvenance #ModelDrift #DataQuality #RecommendationSystems #Ecommerce #DataEngineering #Tech #Business #FexingoBusiness #BusinessPodcast #DataSciencePodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode of The Data Science Podcast, Lucas and Luna explore how knowledge graphs are transforming recommendation engines beyond collaborative filtering. They dive into a real example: how a music streaming service uses knowledge graphs to recommend obscure tracks that collaborative filtering would miss. The hosts break down the graph structure, the role of graph embeddings, and how hybrid approaches combine the best of both worlds. They also discuss the practical challenges of building and maintaining knowledge graphs at scale, from data quality to compute costs. If you've ever wondered why your recommendations sometimes feel surprisingly good, this episode explains the mechanics behind it. Tune in to understand how knowledge graphs are becoming a key tool in the modern data scientist's toolkit.

#KnowledgeGraphs #RecommendationEngines #GraphEmbeddings #DataScience #MachineLearning #CollaborativeFiltering #GraphNeuralNetworks #DataEngineering #MusicStreaming #Tech #DataDriven #FexingoBusiness #BusinessPodcast #Analytics #AI #ML #DataMining #RecommenderSystems

Keep every episode free: buymeacoffee.com/fexingo

Lucas and Luna explore Shapley values, the game-theory concept behind SHAP, which data scientists use to fairly distribute model predictions among features. They break down the math in plain English, show why it's more principled than simple feature importance, and discuss its real-world impact on model fairness and interpretability. With examples from credit scoring and fraud detection, they explain how Shapley values help data teams answer 'why did the model say no?' and why that matters for both compliance and trust. The episode also touches on the computational cost and alternatives like permutation importance, giving listeners a clear picture of when to use each. Perfect for data scientists, ML engineers, and anyone curious about how to make black-box models more transparent.

#ShapleyValues #SHAP #ModelInterpretability #FeatureImportance #FairnessMetrics #GameTheory #CreditScoring #FraudDetection #MachineLearning #DataScience #Technology #AI #ExplainableAI #ModelExplainability #FexingoBusiness #BusinessPodcast #DataDriven #MLFairness

Keep every episode free: buymeacoffee.com/fexingo

On this episode, Lucas and Luna drill into anomaly detection in manufacturing, using the semiconductor industry as a case study. They explore how a fab uses sensor data from etching tools to spot subtle process drifts before they ruin a batch of wafers — saving millions in scrap and downtime. The conversation covers the shift from simple threshold alerts to unsupervised and semi-supervised models, the challenge of labeling rare defects, and how domain engineers and data scientists collaborate to keep false positives low. Lucas shares a concrete example of how a vibration sensor on a robotic arm flagged a bearing failure two weeks before a breakdown, and they discuss why reconstruction error in autoencoders often beats classic statistical methods for high-dimensional sensor data. They also touch on the business side: how to quantify ROI and sell the project to plant managers. The episode closes with thoughts on the future of anomaly detection, from foundation models to edge deployment. If you're a data scientist in manufacturing or curious about how ML keeps factories running, this one's for you.

#AnomalyDetection #ManufacturingAI #PredictiveMaintenance #Semiconductor #SensorData #UnsupervisedLearning #Autoencoders #ProcessControl #Industry40 #DataScience #QualityControl #MachineLearning #SmartFactory #ROI #FexingoBusiness #BusinessPodcast #Technology #DataDriven

Keep every episode free: buymeacoffee.com/fexingo

Lucas and Luna explore how data scientists are using embeddings to transform search. From semantic understanding to handling typos and synonyms, they unpack the shift from keyword matching to meaning-based retrieval. With a practical case study on Wikipedia's search overhaul, they discuss the challenges of scaling embeddings, the importance of evaluation, and how generative AI is changing the game. If you've ever wondered what powers the search bar in your favorite app, this episode breaks it down.

#Embeddings #Search #DataScience #MachineLearning #VectorSearch #SemanticSearch #Wikipedia #NLP #Technology #Business #FexingoBusiness #BusinessPodcast #DataPodcast #AI #DeepLearning #RecommendationSystems #InformationRetrieval #DataEngineering

Keep every episode free: buymeacoffee.com/fexingo

Bias in machine learning isn't just a fairness problem — it's a business risk. In this episode, Lucas and Luna unpack how data scientists are using fairness metrics like demographic parity and equalized odds to audit models before they ship. They walk through a real case: a credit approval model that quietly denied loans to applicants from certain postal codes, and how the team caught it with a bias audit. Along the way, they discuss the limits of fairness metrics, the trade-off between accuracy and equity, and why bias is often a data problem, not just an algorithm problem. If you're building ML systems, this episode gives you a practical framework for measuring and mitigating bias — plus the questions you should ask your data team before deploying a model that affects people's lives.

#BiasInAI #FairnessMetrics #MachineLearning #DataScience #CreditModeling #AlgorithmicFairness #DataEthics #ResponsibleAI #AI #ML #Tech #Business #FexingoBusiness #BusinessPodcast #DataSciencePodcast #Fexingo #Technology #Podcast

Keep every episode free: buymeacoffee.com/fexingo

Recommender systems quietly power everything from your streaming queue to your shopping feed, but most teams still retrain them on a fixed schedule. In this episode, Lucas and Luna dig into online learning — the technique that lets models update continuously as new user behavior streams in. They break down the difference between batch retraining and incremental updates, and walk through a real-world case: a mid-size e-commerce company that cut its model update lag from once a night to every few minutes. The hosts explain the core math — stochastic gradient descent updates on mini-batches — and the practical tradeoffs: more frequent updates can mean more data drift, so you need careful monitoring and rollback strategies. They also touch on the infrastructure shift, from scheduled Spark jobs to streaming pipelines like Kafka and Flink. By the end, you'll understand why online learning is becoming the default for personalization teams, and how to decide if it's right for your system. No jargon for jargon's sake — just the concrete decisions data scientists face when they move from batch to real time.

#OnlineLearning #RecommendationSystems #DataScience #MachineLearning #RealTimeML #StreamingData #ModelRetraining #StochasticGradientDescent #Personalization #Ecommerce #DataDrift #ModelMonitoring #Kafka #Flink #Mlops #Technology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode, Lucas and Luna unpack the quiet killer of machine learning systems: data drift. They trace how models trained on 2024 data start stumbling by 2026, using a concrete example from retail demand forecasting where a model quietly degraded over eighteen months. The conversation walks through the difference between covariate shift and concept drift, why traditional retraining schedules often miss the real problem, and how modern ML teams use statistical tests like the Kolmogorov-Smirnov test and population stability index to catch drift before it hits the bottom line. They also get into the practical side: what a drift dashboard actually looks like, who owns the alert response, and why feature stores with built-in drift detection are becoming the new standard. If you build or manage ML systems, this episode gives you a clear mental model for why models age and what you can do about it.

#DataDrift #ConceptDrift #CovariateShift #MachineLearning #ModelMonitoring #MLOps #FeatureStores #DataScience #RetailAnalytics #KolmogorovSmirnov #PopulationStabilityIndex #DriftDetection #ModelRetraining #DataQuality #Technology #FexingoBusiness #BusinessPodcast #DataPodcast

Keep every episode free: buymeacoffee.com/fexingo