Learn how to estimate the performance of ML models in production and figure out if your model performance is degrading. — In this tutorial, we will use the NYC Green Taxi Dataset to build a machine-learning model that predicts the tip amount a passenger will leave after a taxi ride. Later, we will use NannyML to monitor this model and measure its performance with unseen production data. …

The performance of ML models degrades as time passes and data distribution changes. — A recent study from MIT, Harvard, The University of Monterrey, and Cambridge showed that 91% of ML models degrade over time. This study is one of the first of its kind, where researchers focus on studying machine learning models’ behavior after deployment and how their performance evolves with unseen data. …

In this blog post, we will learn from Zillow’s incident and understand how ML monitoring tools can help you detect the early effects of model performance decay and data drift. In November 2021, the American real-state company Zillow reported a loss of almost $500 M. …

The first time you work with text data is the hardest. You miss having a “simple” tabular dataset, where you can filter and join tables easily. This article will help you cut through your first NLP project and end it with great results. We will be working with tweets data…

Here is a list with 10 of the most common data science interview questions for junior positions. Interviewers usually focus on Statistics fundamentals and Machine Learning concepts. Q1. What two parameters define a normal distribution? Answer: It’s mean and it’s standard deviation. Q2. What is One Hot Encoding? Answer: The process to encode/transform data into an sparse vector in which one element is set…

Spoiler alert I learned nothing. Every interviewer was nice and friendly. But most of the interviews had these awkward silences that I assume are normal(?). As an introvert, I never know how to fill these gaps in the conversation. During and after each interview I felt like a fraud. And…

Three years ago (back when teaching RNNs to generate text was cool) I had this idea of training a TensorFlow model with Harry Potter’s books. This project was on my list and it was hunting me. This weekend I finally did it and also made a simple Streamlit app for…

Un día de estos me puse a pensar por dónde empezaría a aprender machine learning y data science si tuviera que comenzar de nuevo. Y lo vacilón de este ejercicio es que la ruta que me imaginé no fue nada similar a la que hice en su momento. Estoy consciente…

The other day I was taking a shower and a thought came to me: How many times can I exchange a dollar to Colones (Costa Rican currency) then back to dollars and then to colones and so on until I have no money left? I got out the shower and…