Udacity Machine Learning Nanodegree Udacity Connect Intensive Cheatsheet Key Concepts

• Prerequisites
• Basic Statistics
• Python coding
• Intro to Linear Algebra
• Basic Statistics
• Feature versus Target
• Features are data attributes, variables, we use for training in order to predict the target result. 
• Training versus testing
• Split data into training and testing subsets and shuffle
• Train test split
• Data Split
• Further reading http://www.win-vector.com/blog/2015/01/random-testtrain-split-is-not-always-enough/
• Performance Metric
• Coefficient of Determination R2
• http://stattrek.com/statistics/dictionary.aspx?definition=coefficient_of_determination
• Decision Tree and maximum depth
• Not covered in details online
• We had a separate lecture in-person on decision tree
• Entropy - coin example
• Coin is random, 50% head 50% tail, can't predict it Entropy = 1 bit
• Coin with double heads, 100% head, can predict always get head, no Entropy. Entropy = 0
• Learning Curve
• Model Selection
• occam's razor the simpler model is preferred [wikipedia source https://en.wikipedia.org/wiki/Occam%27s_razor]