Hi,
I am trying to interpret a MARs model for classification.
Below is an example taken from the recipes package where I'm trying to predict the factor Status
Taking the field Expenses as an example, how do I read/interpret the co-efficent attached to Expenses-113 - 0.0102 and where in my diagram can I see this? Is this everything to the right of my red vertical line (set at 113).
The next question is related to the intercept. If my model below the intercept is set to 0.985
Does this mean on average someone in my dataset is 98.5% likely to be a good customer?
In that case are the co-efficents in reference to the intercept or am I talking out of my hat 
Thank you very much for your time. Have a nice weekend
suppressPackageStartupMessages(library(tidyverse))
#> Warning: package 'tidyverse' was built under R version 3.5.2
#> Warning: package 'ggplot2' was built under R version 3.5.3
#> Warning: package 'tibble' was built under R version 3.5.2
#> Warning: package 'readr' was built under R version 3.5.2
#> Warning: package 'purrr' was built under R version 3.5.2
#> Warning: package 'dplyr' was built under R version 3.5.2
#> Warning: package 'stringr' was built under R version 3.5.2
#> Warning: package 'forcats' was built under R version 3.5.2
suppressPackageStartupMessages(library(earth))
#> Warning: package 'earth' was built under R version 3.5.2
#> Warning: package 'plotmo' was built under R version 3.5.2
#> Warning: package 'plotrix' was built under R version 3.5.2
suppressPackageStartupMessages(library(recipes))
#> Warning: package 'recipes' was built under R version 3.5.1
mydf <- credit_data %>%
na.omit()
table(mydf$Status)
#>
#> bad good
#> 1026 3013
mars_mod <- earth(Status~., mydf)
coefficients(mars_mod) %>%
enframe() %>%
arrange(desc(abs(value)))
#> # A tibble: 19 x 2
#> name value
#> <chr> <dbl>
#> 1 (Intercept) 0.985
#> 2 Recordsyes -0.283
#> 3 Jobpartime -0.243
#> 4 Maritalseparated -0.159
#> 5 Homepriv -0.0838
#> 6 Homeparents 0.0618
#> 7 h(4-Seniority) -0.0328
#> 8 h(Expenses-113) 0.0102
#> 9 h(Seniority-18) -0.00984
#> 10 h(Seniority-4) 0.00840
#> 11 h(Expenses-75) -0.00375
#> 12 h(130-Income) -0.00319
#> 13 h(Age-19) -0.00137
#> 14 h(113-Expenses) 0.00133
#> 15 h(Amount-475) -0.000276
#> 16 h(1175-Price) -0.000221
#> 17 h(Price-1175) 0.0000745
#> 18 h(2432-Debt) 0.0000368
#> 19 h(5500-Assets) -0.0000291
# Now we build up the PDP Plots
yhat_earth <- function(object, newdata) as.numeric(predict(object, newdata, type = "response"))
# TO confirm we are predicting the Yes label
predict(mars_mod, mydf, type = "response") %>% head(10)
#> good
#> [1,] 0.8289485
#> [2,] 0.9477466
#> [3,] 0.4210298
#> [4,] 0.8097659
#> [5,] 0.7307951
#> [6,] 0.9320309
#> [7,] 0.9131152
#> [8,] 0.9330565
#> [9,] 0.6851392
#> [10,] 0.2086033
# Expenses-113 - 0.0102, 113-Expenses - 0.00133
pdp::partial(mars_mod, pred.var = "Expenses", grid.resolution = 50) %>%
autoplot() +
geom_vline(xintercept = 113, colour = "red")

Created on 2019-09-27 by the reprex package (v0.2.1)