РЕАЛЬНОЕ собеседование на Data Scientist. Алгоритмы + Python

A real Data Science interview at Lamoda with an offer of 400,000 rubles. I've analyzed all the mistakes the candidate made during the interview. 🐳 IT job search guide with an offer of 200,000 rubles in 3-4 months: https://t.me/gernar228/788 (get it, it's free) 🔑 Fill out the application on the website and get a consultation/mock interview with a mentor: https://clck.ru/3REyur (official website for "Turnkey Offer") TIMESTUDY: 00:00 — What's going on now? 00:17 — How to find the median in SQL 03:17 — How a dictionary works in Python and the complexity of key access 03:43 — What is a hash map and what properties should it have 05:12 - a really cool guide, check it out 05:39 — Mutable and immutable data types in Python 06:08 — Types of sorting and the idea of ​​mergesort 06:51 — p-value and elements of statistical testing 11:51 — A problem!! Two white balls from a bag 12:50 — ROC-AUC: probabilistic interpretation 15:00 - How is a decision tree constructed? 16:10 — How does a random forest work? 17:55 - How to get a 200k+ offer in DS? 18:40 — What else is sampled in Random Forest besides strings? 19:04 — How boosting works in general 19:35 — What trees learn in boosting and where is the gradient calculated 22:43 — What happens if you remove the first tree from the ensemble 23:57 — Random Forest Predictions 24:52 — How to reduce the number of CNN parameters without losing the receptive field 26:00 — How to scale a small image dataset without spending money 26:36 — What happens if all network weights are initialized identically 26:57 — How to sort a very large string without additional memory 28:21 — What is left-anti join in Spark 29:05 — Top 5 daily SQL products by speed 31:51 — Bias and Variance: Random Forest vs. Boosting 33:08 — Strong Inference