LO 4.2.2.B

Learning Objective: Define l2 norm and scale equivalent.

Review:

       

<math xmlns="http://www.w3.org/1998/Math/MathML"><msub><mfenced open="|" close="|"><mfenced open="|" close="|"><mi>&#x3B2;</mi></mfenced></mfenced><mn>2</mn></msub><mo>=</mo><msqrt><munderover><mo>&#x2211;</mo><mrow><mi>j</mi><mo>=</mo><mn>1</mn></mrow><mi>p</mi></munderover><msubsup><mi>&#x3B2;</mi><mi>j</mi><mn>2</mn></msubsup></msqrt><mo>=</mo><msqrt><msubsup><mi>&#x3B2;</mi><mn>1</mn><mn>2</mn></msubsup><mo>+</mo><msubsup><mi>&#x3B2;</mi><mn>2</mn><mn>2</mn></msubsup><mo>+</mo><mo>,</mo><mo>.</mo><mo>.</mo><mo>.</mo><mo>.</mo><mo>+</mo><msubsup><mi>&#x3B2;</mi><mi>p</mi><mn>2</mn></msubsup></msqrt></math>

The estimated coefficients  𝛽j of the standard least-squares regression are scale equivariant/equivalent, that is,  multiplying Xj by a constant c simply leads to a scaling of the least-squares coefficient estimates 𝛽j by a factor of 1/c.