Progress, it turned out, was a constant battle. Two steps forward followed by one lingering thought that you didn’t deserve to take those steps in the first place. But maybe the progress and doubt didn’t have to be opposing forces. Maybe they could coexist. And maybe, they could even help each other.
Like many people, I’ve been entertaining myself at home by baking a ton. I’ve gotten pretty good at following recipes, but I decided I wanted to take things one step further and understand the science behind what differentiates a cake from a bread or a cookie. I also like machine learning so I thought: what if I could combine it with baking??!
“You are the coolest person I've ever met on a plane,” he messaged me immediately after I sat down. “There's an empty row behind me. Come say hi!”
Anomaly detection can be a good candidate for machine learning, since it is often hard to write a series of rule-based statements to identify outliers in data. In this post I'll ook at building a model for fraud detection on financial data. If you're thinking *groan, that sounds boring*, don't go away just yet! Fraud detection addresses some interesting challenges in ML.
Today is my five year Googleversary! I thought I'd take a few minutes to share some things I've learned over the past five years related to developer advocacy, working in tech, and other random thoughts.
Learn how to use custom containers on Cloud AI Platform to train an XGBoost model with automated hyperparameter tuning.
In this post we'll walk through how to interpret an XGBoost model trained on a mortgage dataset using the What-if Tool, SHAP, and Cloud AI Platform.
I collected my own sleep data using an Oura ring and analyzed it with BigQuery.
Learn how to build a bag of words text classification model and interpret the model's output with SHAP.
With all the tools democratizing machine learning these days, it’s easier than ever to build high accuracy machine learning models. But even if you build a model yourself using an open source framework like TensorFlow or Scikit Learn, it’s still mostly a black box - it’s hard to know exactly why your model made the prediction it did. As model builders, we’re responsible for the predictions generated by our models and being able to explain...