在R中存储/保存预测模型

Question

I would like to store the output of the following code. I tried list, data frame and vector but was not able to store it.

Model<-lda( y ~ Trend+Class+F1+F4+ATR+macd_signal+macd1+F5 ,data=x)


> Model

Call:
lda(y ~ Trend + Class + F1 + F4 + ATR + macd_signal + macd1 + 
    F5, data = x)

Prior probabilities of groups:

     LOSS    PROFIT 
0.4981818 0.5018182 

Group means:

       TrendBull   ClassUP       F1       F4      ATR   macd_signal       macd1       F5
LOSS   0.5450122 0.3990268 1480.451 1481.672 11.64657 -0.0005850151  0.01542818 1478.567
PROFIT 0.5000000 0.4082126 1487.280 1486.707 12.25799 -0.0304256947 -0.03845741 1489.620

Coefficients of linear discriminants:

                     LD1
TrendBull   -0.033267160

ClassUP      0.151291378

F1          -0.003215276

F4          -0.042431558

ATR          0.082615338

macd_signal  0.090182055

macd1        2.637216918

F5           0.045956343

> class(Model)

[1] "lda"

> typeof(Model)

[1] "list"

Any ideas on how to store it?

Answer 1

One thing you can do is save the model output in its own file. An R object can be saved as an rds file. For example:

saveRDS(model, file = "your path/filename.rds")

You can then load the object into your R environment:

mod <- readRDS("your path/filename.rds")

Answer 2

TL;DR: Model is itself a list that contains all of the desired information and more. Run str(Model) to see its components. However, you can also extract just the information you want and save it in a list as follows:

my_lda_smry = function(x) {
  list(Call=x$call, Prior=x$prior, `Group Means`=x$means, 
     Coefficients=x$scaling,
     `Proportion of Trace`= round(x$svd^2/sum(x$svd^2), 4))
}

Model_summary = my_lda_smry(Model)

See below for additional explanation.

---

---

Model is an lda model object, which is a list containing all the output of the lda function. When you type Model , R is calling the print "method" for that lda model object, which is a function called print.lda (typing Model in the console is equivalent to typing print(Model) , which runs the print.lda function on your model object). So, you can look at the model object and the print.lda function to see what they're doing and then create your own summary object from that.

Look at code for print.lda

The code for print.lda is below. Note that print.lda is getting each of its outputs from the model object. For example x$prior is the prior probabilities of each group. We just need to create a function to extract each of elements we want.

getAnywhere(print.lda)

 function (x, ...) { if (!is.null(cl <- x$call)) { names(cl)[2L] <- "" cat("Call:\\n") dput(cl, control = NULL) } cat("\\nPrior probabilities of groups:\\n") print(x$prior, ...) cat("\\nGroup means:\\n") print(x$means, ...) cat("\\nCoefficients of linear discriminants:\\n") print(x$scaling, ...) svd <- x$svd names(svd) <- dimnames(x$scaling)[[2L]] if (length(svd) > 1L) { cat("\\nProportion of trace:\\n") print(round(svd^2/sum(svd^2), 4L), ...) } invisible(x) } 

Look at the model object returned by lda

Now let's create a model and look at the model object. We don't actually need to do look at the model object here, because print.lda tells us what we need to know. However, knowing the structure of the model object can be helpful if you want to extract information that isn't returned by the standard extractor functions provided with a package. Note the the model object is a list containing various types of information about the model.

library(MASS)

model = lda(mpg ~ wt + hp + carb + cyl, data=mtcars)

str(model)

 List of 10 $ prior : Named num [1:25] 0.0625 0.0312 0.0312 0.0312 0.0312 ... ..- attr(*, "names")= chr [1:25] "10.4" "13.3" "14.3" "14.7" ... $ counts : Named int [1:25] 2 1 1 1 1 2 1 1 1 1 ... ..- attr(*, "names")= chr [1:25] "10.4" "13.3" "14.3" "14.7" ... $ means : num [1:25, 1:4] 5.34 3.84 3.57 5.34 3.57 ... ..- attr(*, "dimnames")=List of 2 .. ..$ : chr [1:25] "10.4" "13.3" "14.3" "14.7" ... .. ..$ : chr [1:4] "wt" "hp" "carb" "cyl" $ scaling: num [1:4, 1:4] 4.668 -0.0115 -3.6744 -3.8415 4.2625 ... ..- attr(*, "dimnames")=List of 2 .. ..$ : chr [1:4] "wt" "hp" "carb" "cyl" .. ..$ : chr [1:4] "LD1" "LD2" "LD3" "LD4" $ lev : chr [1:25] "10.4" "13.3" "14.3" "14.7" ... $ svd : num [1:4] 10.51 3.42 1.49 1.05 $ N : int 32 $ call : language lda(formula = mpg ~ wt + hp + carb + cyl, data = mtcars) $ terms :Classes 'terms', 'formula' language mpg ~ wt + hp + carb + cyl .. ..- attr(*, "variables")= language list(mpg, wt, hp, carb, cyl) .. ..- attr(*, "factors")= int [1:5, 1:4] 0 1 0 0 0 0 0 1 0 0 ... .. .. ..- attr(*, "dimnames")=List of 2 .. .. .. ..$ : chr [1:5] "mpg" "wt" "hp" "carb" ... .. .. .. ..$ : chr [1:4] "wt" "hp" "carb" "cyl" .. ..- attr(*, "term.labels")= chr [1:4] "wt" "hp" "carb" "cyl" .. ..- attr(*, "order")= int [1:4] 1 1 1 1 .. ..- attr(*, "intercept")= int 1 .. ..- attr(*, "response")= int 1 .. ..- attr(*, ".Environment")=<environment: R_GlobalEnv> .. ..- attr(*, "predvars")= language list(mpg, wt, hp, carb, cyl) .. ..- attr(*, "dataClasses")= Named chr [1:5] "numeric" "numeric" "numeric" "numeric" ... .. .. ..- attr(*, "names")= chr [1:5] "mpg" "wt" "hp" "carb" ... $ xlevels: Named list() - attr(*, "class")= chr "lda" 

Extract the information we want and return it in a list

The function below just mimics what print.lda does, but saves the results in a list, rather than printing it to the screen:

my_lda_smry = function(x) {
  list(Call=x$call, Prior=x$prior, `Group Means`=x$means, 
     Coefficients=x$scaling,
     `Proportion of Trace`= round(x$svd^2/sum(x$svd^2), 4))
}

Now run the function:

m.smry = my_lda_smry(model)

 $Call lda(formula = mpg ~ wt + hp + carb + cyl, data = mtcars) $Prior 10.4 13.3 14.3 14.7 15 15.2 15.5 15.8 16.4 17.3 17.8 18.1 18.7 19.2 19.7 21 0.06250 0.03125 0.03125 0.03125 0.03125 0.06250 0.03125 0.03125 0.03125 0.03125 0.03125 0.03125 0.03125 0.06250 0.03125 0.06250 21.4 21.5 22.8 24.4 26 27.3 30.4 32.4 33.9 0.06250 0.03125 0.06250 0.03125 0.03125 0.03125 0.06250 0.03125 0.03125 $`Group Means` wt hp carb cyl 10.4 5.3370 210.0 4.0 8 13.3 3.8400 245.0 4.0 8 14.3 3.5700 245.0 4.0 8 14.7 5.3450 230.0 4.0 8 15 3.5700 335.0 8.0 8 15.2 3.6075 165.0 2.5 8 15.5 3.5200 150.0 2.0 8 15.8 3.1700 264.0 4.0 8 16.4 4.0700 180.0 3.0 8 17.3 3.7300 180.0 3.0 8 17.8 3.4400 123.0 4.0 6 18.1 3.4600 105.0 1.0 6 18.7 3.4400 175.0 2.0 8 19.2 3.6425 149.0 3.0 7 19.7 2.7700 175.0 6.0 6 21 2.7475 110.0 4.0 6 21.4 2.9975 109.5 1.5 5 21.5 2.4650 97.0 1.0 4 22.8 2.7350 94.0 1.5 4 24.4 3.1900 62.0 2.0 4 26 2.1400 91.0 2.0 4 27.3 1.9350 66.0 1.0 4 30.4 1.5640 82.5 2.0 4 32.4 2.2000 66.0 1.0 4 33.9 1.8350 65.0 1.0 4 $Coefficients LD1 LD2 LD3 LD4 wt 4.66796895 4.262520788 0.35307402 -0.67013561 hp -0.01149489 0.005714994 0.04376624 0.01627358 carb -3.67441417 -0.581458148 -0.44870373 -0.83522067 cyl -3.84149993 -0.911662765 -1.52258858 0.37350681 $`Proportion of Trace` [1] 0.8804 0.0930 0.0178 0.0088