Uplift-археология: как откопать эффект в слоях шума
В uplift‑моделировании выбор алгоритма — лишь часть задачи. Результат во многом зависит от постановки эксперимента, качества исторических данных, схемы валидации, выбора метрик и,...
В uplift‑моделировании выбор алгоритма — лишь часть задачи. Результат во многом зависит от постановки эксперимента, качества исторических данных, схемы валидации, выбора метрик и,...
Generative AI makes it cheap to produce personalized content at scale, but which variation do you show each customer? Amazon Payments used a multi-objective contextual bandit on Am...
On 1 million simulated customers, calibrated Qini scored a Linear DR learner 0.0856 against 0.0826 for the true-CATE ranking that should have won. The post When Qini Rewards the Wr...
NOTE: Any updates to this article will be maintained here: https://medley.bryer.org/articles/stratified_glm.html Introduction Statistical modeling when there is substantial imbalan...
A model was already live, forecasting rain a day ahead. It looked done. Evaluating it properly is what showed it wasn’t.Continue reading on Medium »
This week, I worked through Module 3: Machine Learning for Classification in ML Zoomcamp 2026. The goal was to build a model that could help a telecom company identify customers w...
Summary Who this is for: Data scientists and analysts who need to estimate what an intervention does (a call, a discount, a campaign) and want to know when causal ML methods are wo...
Abacus.AI applied the same fine-tuning recipe to three very different open-weight models. The gains landed in completely different places & that tells you more than any single...
Amazon SageMaker HyperPod Data Capture records live inference inputs and outputs to Amazon S3. This post walks through one end-to-end example: capturing a support router's producti...
No training data, no gradients, no fine-tuning run.Continue reading on Medium »
Built by Salesforce AI Research, GIFT-Eval-CTX measures how effectively LLMs and agentic AI convert forward-looking context, such as promotions or weather events, into precise fore...
Most gains in a supervised learning project don't come from switching algorithms; they come from what you feed the algorithm. Feature engineering is the process of transforming raw...
Property Finder found that 96.8% of their vision-AI evaluation labels were inherited from the incumbent model, hiding its true accuracy. See how they built unbiased ground truth, a...
by Manuel Brenner, Ismail Guennouni, Daniel Durstewitz, Stephanie N. L. Schmidt, Stefanie Lis, Peter Kirsch, Georgia Koppe Social decision making is inherently complex, shaped by...
Use SocialBu to discover ideas, generate post drafts, and schedule them across your social channels.