Input → Model → Output
One of the simplest ways to understand a Machine Learning system is to think about three parts: what goes in, what processes it, and what comes out.
Data given to the model.
Uses what was learned from data.
The result produced by the model.
What Is the Input?
The input is the information we give to a trained model when we want it to produce an output.
What the input contains depends on the problem we are trying to solve.
For example, 1,500 square feet.
For example, words and sender information.
What Is the Model?
The model is the part of the system that represents what was learned from training data.
Once trained, the model can take new input and produce an output.
The model is not simply the original training data. It is the result of a learning process applied to that data.
What Is the Output?
The output is the result produced by the model after it receives input.
Predicted house price.
Predicted category.
Recommended item.
The House Price Example
Let's connect the three parts to the house-price problem from Lesson 1.
Information about the new house.
Uses the relationship learned from previous examples.
Predicted house price.
Training Happens Before This Flow
There is an important detail here. The simple Input → Model → Output flow assumes that the model already exists.
After training, we can provide new input to the model and receive an output.
Training data → model
New input → output
Input Does Not Mean "Everything About the Object"
The input contains the information we choose to provide to the model.
A house has many properties, but our simple model might only receive its size.
Size, location, age, bedrooms, condition, and more.
Only the information provided to this particular model.
Different Problems Have Different Outputs
Input → Model → Output is a general structure. The meaning of the output depends on the problem.
Example: predict house price.
Example: spam or not spam.
Example: recommend a product.
Input → Model → Output
The input provides information. The trained model applies what it learned. The output is the result produced for that input.
Identify the Three Parts
Input: house size → Model: learned relationship → Output: predicted price.
Input: email information → Model: learned pattern → Output: spam or not spam.