Regression by remove point and check the quality of regression

Question

I want to do a linear regression by removing values ​​to evaluate the performance of the regressions in terms of correlation coefficient in comparison to a threshold value

datas <- data.frame(x= c(0,0.45,2,8,10,18,30,45,50,52),y= c(1,3,5,7,9,17,26,44,48,50))

This is my function and i dont know how can remove first or last point and run and compare the result automatically

  
  formula_lm <- as.formula(paste0(py,"~",px))
  model <- lm(formula_lm, data = df)
  rslt <- df %>%
    mutate(a = round(model$coefficients[2],3),
           r_carre = round(summary(model)$r.squared,3),
           comp_val_critic = ifelse(r_carre <=threeshold, "linearity_down","linearity_high"))
  rslt
} ```

Use function

``` lm_f(df = datas,
           py = "y",
           px = "x")```