CODEWITHSUNDEEP
rgbmdismo
F.Johann
F.Johann

Reputation: 25

gbm.interactions for gbm.fixed

How can I get interaction assessment for gbm calculated with fixed n.trees? I tried:

data(Anguilla_train)
angaus.fixed <- gbm.fixed(data=Anguilla_train, gbm.x = 3:13, 
  gbm.y = 2,family = "bernoulli", tree.complexity = 5, learning.rate = 0.01,
  bag.fraction = 0.5, n.trees = 5000)
find.int <- gbm.interactions(angaus.fixed)

But:

1 Error in [.data.frame(pred.frame, , n) : undefined columns selected

Upvotes: 1

Views: 500

Answers (1)

Marco Sandri
Marco Sandri

Reputation: 24262

After a careful examination of the dismo::gbm.interactions function, I found that dismo::gbm.fixed returns an object that is not fully compatible with gbm.interactions.
Here is how your code could be modified:

library(dismo)  
data(Anguilla_train)
angaus.fixed <- gbm.fixed(data=Anguilla_train, gbm.x = 3:13, 
  gbm.y = 2,family = "bernoulli", tree.complexity = 5, learning.rate = 0.01,
  bag.fraction = 0.5, n.trees = 5000)

# Change the name of angaus.fixed$gbm.call$data with angaus.fixed$gbm.call$dataframe
names(angaus.fixed$gbm.call)[1] <- "dataframe"

find.int <- gbm.interactions(angaus.fixed)

######
gbm.interactions - version 2.9
Cross tabulating interactions for gbm model with 11 predictors
1 2 3 4 5 6 7 8 9 10

Now gbm.interactions works nicely and yields the following results:

find.int$interactions

#######
           SegSumT SegTSeas SegLowFlow DSDist DSMaxSlope USAvgT USRainDays USSlope USNative DSDam Method
SegSumT          0    28.35       0.88 150.10      13.20  24.38      30.30   17.85    22.57  0.27  44.68
SegTSeas         0     0.00      17.66 130.69       7.96  16.93       3.14   18.59    25.90  0.14   4.15
SegLowFlow       0     0.00       0.00   5.92       3.85   6.88       4.69   10.85     5.06  0.09   6.23
DSDist           0     0.00       0.00   0.00       2.40  10.68      47.77   41.54     6.82  0.01  11.42
DSMaxSlope       0     0.00       0.00   0.00       0.00   3.22       2.85    3.80     2.64  0.00   1.25
USAvgT           0     0.00       0.00   0.00       0.00   0.00       3.08   11.76    24.90  0.04   1.94
USRainDays       0     0.00       0.00   0.00       0.00   0.00       0.00    4.69    15.26  0.09  18.16
USSlope          0     0.00       0.00   0.00       0.00   0.00       0.00    0.00    12.08  0.03  19.44
USNative         0     0.00       0.00   0.00       0.00   0.00       0.00    0.00     0.00  0.02  12.79
DSDam            0     0.00       0.00   0.00       0.00   0.00       0.00    0.00     0.00  0.00   0.13
Method           0     0.00       0.00   0.00       0.00   0.00       0.00    0.00     0.00  0.00   0.00


find.int$rank.list

#######
  var1.index var1.names var2.index var2.names int.size
1          4     DSDist          1    SegSumT   150.10
2          4     DSDist          2   SegTSeas   130.69
3          7 USRainDays          4     DSDist    47.77
4         11     Method          1    SegSumT    44.68
5          8    USSlope          4     DSDist    41.54
6          7 USRainDays          1    SegSumT    30.30

Upvotes: 1

Related Questions