Stay organized with collectionsSave and categorize content based on your preferences.
Creates a reducer that computes a ridge regression with numX independent variables (not including constant) followed by numY dependent variables. Ridge regression is a form of Tikhonov regularization which shrinks the regression coefficients by imposing a penalty on their size. With this implementation of ridge regression there NO NEED to include a constant value for bias.
The first output is a coefficients array with dimensions (numX + 1, numY); each column contains the coefficients for the corresponding dependent variable plus the intercept for the dependent variable in the last column. Additional outputs are a vector of the root mean square of the residuals of each dependent variable and a vector of p-values for each dependent variable. Outputs are null if the system is underdetermined, e.g., the number of inputs is less than numX + 1.
Usage
Returns
ee.Reducer.ridgeRegression(numX,numY,lambda)
Reducer
Argument
Type
Details
numX
Integer
the number of independent variables being regressed.
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Missing the information I need","missingTheInformationINeed","thumb-down"],["Too complicated / too many steps","tooComplicatedTooManySteps","thumb-down"],["Out of date","outOfDate","thumb-down"],["Samples / code issue","samplesCodeIssue","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2024-07-13 UTC."],[[["\u003cp\u003eCreates a reducer for ridge regression, a regularization method that shrinks regression coefficients to prevent overfitting.\u003c/p\u003e\n"],["\u003cp\u003eOutputs include regression coefficients, root mean square of residuals, and p-values for each dependent variable.\u003c/p\u003e\n"],["\u003cp\u003eRequires specifying the number of independent and dependent variables, along with an optional regularization parameter (lambda).\u003c/p\u003e\n"],["\u003cp\u003eThe reducer automatically handles the intercept term, so there's no need to add a constant value for bias.\u003c/p\u003e\n"],["\u003cp\u003eOutputs will be null if the system is underdetermined, meaning there are fewer input data points than independent variables plus one.\u003c/p\u003e\n"]]],[],null,["# ee.Reducer.ridgeRegression\n\nCreates a reducer that computes a ridge regression with numX independent variables (not including constant) followed by numY dependent variables. Ridge regression is a form of Tikhonov regularization which shrinks the regression coefficients by imposing a penalty on their size. With this implementation of ridge regression there NO NEED to include a constant value for bias.\n\n\u003cbr /\u003e\n\nThe first output is a coefficients array with dimensions (numX + 1, numY); each column contains the coefficients for the corresponding dependent variable plus the intercept for the dependent variable in the last column. Additional outputs are a vector of the root mean square of the residuals of each dependent variable and a vector of p-values for each dependent variable. Outputs are null if the system is underdetermined, e.g., the number of inputs is less than numX + 1.\n\n| Usage | Returns |\n|-----------------------------------------------------------|---------|\n| `ee.Reducer.ridgeRegression(numX, `*numY* `, `*lambda*`)` | Reducer |\n\n| Argument | Type | Details |\n|----------|---------------------|------------------------------------------------------|\n| `numX` | Integer | the number of independent variables being regressed. |\n| `numY` | Integer, default: 1 | the number of dependent variables. |\n| `lambda` | Float, default: 0.1 | Regularization parameter. |"]]
