Keeping It Classy
ABSTRACT
Classification is one of the core workloads in machine learning and a natural starting point for budding data scientists. Humans have a natural inclination to classify things--that cloud looks like a tree!--but when we apply this to computers, we need to apply a bit of rigor. This talk supplies that bit of rigor, covering the necessary background to solving a classification problem. We will define key terminology, review some of the most popular and effective classification algorithms available today, and explain the aptly-named confusion matrix along the way. Examples will be primarily in Python, although no prior knowledge of the language will be necessary for this session.
ADDITIONAL MEDIA
I have put together a YouTube playlist covering the topics in this talk.
SLIDES
Click here to access the slides for this presentation.
The slides are licensed under Creative Commons Attribution-ShareAlike.
DEMO CODE
Click here to access demo code for this presentation.
The source code is licensed under the terms offered by the GPL.
LINKS & FURTHER INFO
Helpful Resources
- A review of common classification algorithms
- Another review of common classification algorithms
- Yet another review of common classification algorithms
- A tutorial of the scikit-learn classifiers
- The CART algorithm
- Another take on the CART algorithm
- The random forest ensembling algorithm
- An explanation of how bagging (bootstrap aggregating) works
- How XGBoost differs from gradient boosting in scikit-learn
- The K-Nearest Neighbors algorithm
- An overview of Passive-Aggressive classifiers
- The original Passive-Aggressive algorithms paper
- Passive-Aggressive algorithms
- Class imbalance in Passive-Aggressive algorithms
- Understanding the confusion matrix
- Classification model evaluation