[ServeIt](https://github.com/rtlee9/serveit) is an open source library that lets you easily serve model predictions and supplementary information from a RESTful API on any domain using your favorite Python ML library. This post illustrates the process of deploying a pre-trained ImageNet classifier with ServeIt to a new API. The classifier accepts an... [Read More]

Kaggle is a popular online platform for data science competitions. Competition hosts can post their data online, and data scientist from around the world compete to build the best algorithm for the host's needs. Competition hosts benefit when they receive a large volume of high-quality submissions and walk away from... [Read More]

Machine learning models are frequently used to make predictions about a target variable, such as the sales price of a home. However, we can use a fitted model to do much more than to predict -- by carefully inspecting a model we can glean useful insights that can be used... [Read More]

In this post I introduce the _Company Vector_, a distributed representation of a company's core business activities. Company vectors are learned from business descriptions found in public SEC 10-K filings using an unsupervised learning algorithm called the _Paragraph Vector_. The learned vectors can be used to compare companies to each... [Read More]