Function that builds an arid_linreg class model object that provides sklearn linear regression interface functionality and attributes. The arid_linreg function instantiates a linear regression model type based on the input specifications and provides methods to fit/predict/score the results and retrieve the calculated sklearn coefficients.

Function that builds an arid_linreg class model object that provides sklearn linear regression interface functionality and attributes. The arid_linreg function instantiates a linear regression model type based on the input specifications and provides methods to fit/predict/score the results and retrieve the calculated sklearn coefficients.

arid_linreg(regularization = NULL, lambda = NULL)

Arguments

regularization	(character): A string defining NULL,'L1','L2', or 'L1L2' linear coefficient regularization (default: NULL)
lambda	(double):the numeric regularization strength value

Value

arid_linreg: an arid_linreg class linear regression model object

Examples

model <- arid_linreg()
model$fit(matrix(1:12, nrow = 3, ncol = 3), as.matrix(c(1,2,0)))
#> $thisEnv
#> <environment: 0x0000000044d1d950>
#> 
#> $fit
#> function (X, y) 
#> {
#>     if (is.null(X) | length(X) < 1) {
#>         stop("ERROR: Invalid input X feature values to fit")
#>     }
#>     if (is.null(y) | length(y) < 1) {
#>         stop("ERROR: Invalid input y response value to fit")
#>     }
#>     if (is.list(X)) {
#>         if (length(dplyr::select_if(X, is.numeric) != length(X))) {
#>             warning("WARNING: Dropping non-numeric input features in X")
#>             X <- X %>% dplyr::select_if(is.numeric)
#>         }
#>         X <- data.matrix(X)
#>     }
#>     if (is.list(y)) {
#>         warning("WARNING: Input y to fit is a list, converting to matrix")
#>         y <- as.matrix(y)
#>     }
#>     if (!is.numeric(y)) {
#>         stop("ERROR: Response y is not numeric")
#>     }
#>     if (nrow(X) != nrow(y)) {
#>         stop("ERROR: Input samples X and responses y not the same length")
#>     }
#>     model <- NULL
#>     if (is.null(regularization_)) {
#>         lambda_ <- 0
#>         model <- glmnet::glmnet(X, y, family = "gaussian", alpha = 1, 
#>             lambda = lambda_)
#>     }
#>     else if (regularization == c("L1")) {
#>         model <- glmnet::glmnet(X, y, alpha = 1, family = "gaussian", 
#>             lambda = lambda_)
#>     }
#>     else if (regularization == c("L2")) {
#>         model <- glmnet::glmnet(X, y, alpha = 0, family = "gaussian", 
#>             lambda = lambda_)
#>     }
#>     else if (regularization == c("L1L2")) {
#>         model <- glmnet::glmnet(X, y, alpha = 0.5, family = "gaussian", 
#>             lambda = lambda_)
#>     }
#>     coefs <- .get_coefs(X, y, model, lambda_)
#>     arid_linreg$intercept_ <- coefs[1]
#>     arid_linreg$coef_ <- coefs[c(-1)]
#>     assign("coef_", coefs[c(-1)], thisEnv)
#>     assign("model_", model, thisEnv)
#>     return(arid_linreg)
#> }
#> <bytecode: 0x0000000044d231a8>
#> <environment: 0x0000000044d1d950>
#> 
#> $predict
#> function (newx) 
#> {
#>     if (is.null(newx) | length(newx) < 1) {
#>         stop("ERROR: Invalid input new X sample values to predict")
#>     }
#>     if (is.null(model_)) {
#>         stop("ERROR: Must fit model before predicting")
#>     }
#>     if (length(coef_) != ncol(newx)) {
#>         stop("ERROR: Incorrect number of features in newx samples")
#>     }
#>     return(glmnet::predict.glmnet(model_, s = lambda_, newx = newx))
#> }
#> <bytecode: 0x0000000044d1c450>
#> <environment: 0x0000000044d1d950>
#> 
#> $score
#> function () 
#> {
#>     if (is.null(model_)) {
#>         stop("ERROR: Must fit model before scoring")
#>     }
#>     max(model_$dev.ratio)
#> }
#> <bytecode: 0x0000000044d1d2f8>
#> <environment: 0x0000000044d1d950>
#> 
#> $intercept_
#> [1] 2
#> 
#> $coef_
#> [1] -5.000000e-01 -3.330669e-16 -1.232595e-31
#> 
#> attr(,"class")
#> [1] "arid_linreg"
model$predict(t(as.matrix(c(1,2,2))))
#>        1
#> [1,] 1.5
model$score()
#> [1] 0.25