Unit 1: Machine Learning in a Nutshell

Generalization

the model’s ability to adapt to previously unseen data

label

-target variable; attribute you are trying to represent
-represented by y

CODE: {y = df[‘colName’] }

feature

-input for that prediction
-often grouped together as a vector; represented by X

CODE: { X = df[feature_list] } { X = df.drop(columns=’y_colName’ }

supervised learning

-learnt by using prior knowledge
-attempts to discover the relationship between features and an associated label for the purpose of future prediction.

unsupervised learning

discovering patterns in data without the use of training data containing labeled examples; aka on its own!

• first step in ML pipeline

regression

the label is any real valued number

classification

the label is a categorical variable

clustering

• unsupervised learning technique

• group subsets of data that are collectively similar to each other based on the similarity of their feature value

The ML Process

Business Understanding

defining the business objectives

• what does the business need?

example: ____

data understanding and preparation

transforming raw data into a form of suitable for modeling

• what data do we have? is the data clean?

• how do we prepare the data for our model?

example: ______

modeling

• what techniques should we apply to the model?

evaluation

training a model and confirmed its performance on unseen data, perform additional analysis

• is our model best suited for our problem?

example:

deployment

preparing to move a model into production after evaluating it

• how do we make our model available to stakeholders and other users?

example: