Home runs and launch variables

\[ \log(\frac{\text{Pr}(HR)}{1-\text{Pr}(HR)}) = s(LA, LS) \] - Logistic GAM regression of probability of a Home Run vs:

• LA : Launch angle

• LS : Launch speed

• s is used in GAM to indicate a smooth function of the two measurements.

library(mgcv)
fit_23 <- gam(
  HR ~ s(launch_angle, launch_speed),
  family = binomial,
  data = sc_2023
)

We can get predictions from the model:

fit_23 |>
  predict(
    newdata = data.frame(
      launch_speed = 105, 
      launch_angle = 25
    ),
    type = "response"
  )

##     1 
## 0.773