Scribd Logo
Upload

Welcome to Scribd!

  • 3 Bias - Var - Tradeoff

    Uploaded by

    Riya Lokhande
    1 views5 pages
    Linear regression models have three sources of error: bias, variance, and irreducible error. There is a bias-variance tradeoff where decreasing model flexibility reduces variance but increases bias, and increasing flexibility does the opposite. The goal is to minimize total expected error, which is the sum of the bias and variance.

    Original Description:

    Original Title

    3 Bias_Var_tradeoff

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    Linear regression models have three sources of error: bias, variance, and irreducible error. There is a bias-variance tradeoff where decreasing model flexibility reduces variance but increases bias, and increasing flexibility does the opposite. The goal is to minimize total expected error, which is the sum of the bias and variance.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    1 views5 pages

    3 Bias - Var - Tradeoff

    Uploaded by

    Riya Lokhande
    Linear regression models have three sources of error: bias, variance, and irreducible error. There is a bias-variance tradeoff where decreasing model flexibility reduces variance but increases bias, and increasing flexibility does the opposite. The goal is to minimize total expected error, which is the sum of the bias and variance.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    of 5

    Linear Regression

    𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑡𝑒𝑠𝑡 𝑒𝑟𝑟𝑜𝑟 = 𝐸 𝐵𝑖𝑎𝑠 + 𝐸 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒 + 𝐸(𝜖)

    𝐸(𝜖)
    Bias-Variance Variance of error, Irreducible

    Trade-Off 𝐸 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒
    Amount by which predicted function will change if we change training dataset

    𝐸 𝐵𝑖𝑎𝑠
    Error due to approximation of complex relationship as a simpler model such as
    linear model

    Start-Tech Academy
    Linear Regression

    Population regression line

    Sample regression line

    Variance

    Less Flexible method

    More Flexible method

    Start-Tech Academy
    Linear Regression

    𝐸 𝐵𝑖𝑎𝑠
    Error due to approximation of complex relationship as a simpler model such as
    linear model

    Bias

    Start-Tech Academy
    Linear Regression

    The Tradeoff
    If we try to decrease one by changing model
    flexibility, other one increases

    Bias-Variance
    Trade-Off

    Bias + Variance
    Bias
    Variance

    Start-Tech Academy

    You might also like