08. Model Inputs: Features vs Samples

AI For Trading C6 L1 A06 Model Inputs V2

Understanding Data in AI Models

AI models rely heavily on data, specifically broken down into:

  • Features: Independent variables or characteristics for the model to learn from (e.g., years of education, age).
  • Samples: Individual instances collected for each feature, often in thousands.

The Data Matrix

  • Columns represent features.
  • Rows are samples.

Interplay of Components

  1. Samples vs. Model Learning:

    • More samples often lead to better learning.
    • However, there's a point of diminishing returns.

  2. Quality Over Quantity in Features:

    • Features need to be relevant and informative.
    • Avoid redundant features to prevent overfitting.

Importance of Data Quality

  • Representative and bias-free samples are essential for effective learning.

Feature Engineering

  • Uses transformation and augmentation to extract meaningful features.

Next Steps

  • Cover data preparation and other optimization strategies, including feature selection techniques.

Select all true statements.
SOLUTION:
  • Input data can be thought of as a 2D matrix where the columns are features and the rows are samples.
  • For a model that tries to predict house prices, if the area in square feet is one of the features, we should NOT include area in square meters as an additional feature.

Next Concept