## ----intro-knitr-opts, include = FALSE----------------------------------------
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.width = 6, fig.height = 4,
  fig.align = "center"
)

## ----setup--------------------------------------------------------------------
library(MetaHunt)
set.seed(1)

## ----simulate-----------------------------------------------------------------
G <- 40; m <- 120; K_true <- 3
x <- seq(0, 1, length.out = G)
basis <- rbind(sin(pi * x), cos(pi * x), x)         # 3 true bases on the grid

W <- data.frame(w1 = rnorm(m), w2 = rnorm(m))       # study-level covariates
beta <- cbind(c(1, -0.8), c(-0.5, 1.2), c(0, 0))
pi_true <- exp(as.matrix(W) %*% beta)
pi_true <- pi_true / rowSums(pi_true)

F_hat <- pi_true %*% basis + matrix(rnorm(m * G, sd = 0.05), m, G)
dim(F_hat)

## ----fit----------------------------------------------------------------------
fit <- metahunt(F_hat, W, K = 3)
fit

## ----plot-bases---------------------------------------------------------------
plot(fit, x_axis = x,
     col = c("#0072B2", "#D55E00", "#009E73"))

## ----predict------------------------------------------------------------------
W_new <- data.frame(w1 = c(0, 1, -1), w2 = c(0, -0.5, 1),
                    row.names = c("baseline", "high w1, low w2", "low w1, high w2"))
f_pred <- predict(fit, newdata = W_new)
dim(f_pred)
oldpar <- par(mar = c(4, 4.5, 3, 1))
matplot(x, t(f_pred), type = "l", lty = 1,
        col = c("#0072B2", "#D55E00", "#009E73"),
        xlab = "x", ylab = expression(tilde(f)(x)),
        main = "Predicted target functions")
legend("topright", legend = rownames(W_new),
       col = 1:3, lty = 1, bty = "n")
par(oldpar)

## ----predict-wrapped----------------------------------------------------------
predict(fit, newdata = W_new, wrapper = mean)