search

trnnick / nnfor

31 stars 15 forks source link

Error in apply(y, 1L, rank) : dim(X) must have a positive length #4

Open samyoung0424 opened 5 years ago

samyoung0424 commented 5 years ago

It seems that if your ts object doesn't populated at the very beginning of your time, the error Error in apply(y, 1L, rank) : dim(X) must have a positive length will be thrown.

like this:

testdata <- ts(data = as.vector(mydata$V4), start = c(2016,3), frequency = 12)
fit <- mlp(testdata)
Error in apply(y, 1L, rank) : dim(X) must have a positive length
trnnick commented 5 years ago

When you say it is not populated, do you mean it has NAs?

senatorcasey commented 5 years ago

I'm getting the same error, but I don't know if it's for the same reason. Here's my time series: `

ts027016 Time Series: Start = c(2017, 5) End = c(2019, 7) Frequency = 52 [1] 2 5 3 6 3 2 0 0 2 3 1 2 2 2 5 2 1 2 2 3 4 0 2 1 4 1 1 [28] 0 0 1 2 0 0 1 1 1 1 1 1 1 5 2 3 1 0 3 3 0 0 0 8 2 0 0 [55] 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 2 [82] 1 0 3 0 1 2 3 5 1 2 4 2 0 0 0 0 0 0 0 0 12 2 6 4 2 1 mlp(ts027016) Error in apply(y, 1L, rank) : dim(X) must have a positive length`

It seems to work fine for this time series: `

ts038178 Time Series: Start = c(2014, 22) End = c(2019, 10) Frequency = 52 [1] 1 2 2 2 0 0 3 2 8 5 11 1 7 4 4 8 2 5 4 3 2 2 2 4 3 1 3 [28] 4 3 4 5 1 2 4 2 4 1 3 2 3 2 6 3 2 1 4 4 5 1 6 2 3 4 3 [55] 7 7 4 6 5 3 9 9 8 9 10 8 11 5 3 4 5 2 2 3 5 5 6 10 7 8 6 [82] 2 8 6 6 8 7 5 5 3 4 4 4 6 4 3 3 4 4 4 7 6 10 14 6 7 11 14 [109] 13 9 21 14 18 18 9 16 19 14 6 17 13 7 10 8 5 10 5 10 9 5 7 5 10 7 0 [136] 6 4 8 10 11 19 9 8 5 8 11 13 10 6 14 13 4 15 17 18 10 14 12 10 14 22 22 [163] 17 28 27 11 22 26 15 18 16 19 14 9 10 5 8 10 7 6 7 7 10 6 10 13 10 10 12 [190] 9 5 14 11 8 11 13 5 6 10 9 7 8 11 9 10 10 12 8 12 9 13 14 14 12 25 15 [217] 23 14 28 14 19 18 8 21 14 18 9 6 10 19 7 4 15 13 2 3 6 6 15 13 7 14 13 [244] 15 11 7 12 1 7 mlp(ts038178) MLP fit with 5 hidden nodes and 20 repetitions. Series modelled in differences: D1D52. Univariate lags: (1,2,3,4,5,6,11,12,14,16,20,22,24,26,39,44,45,47,48,52) Forecast combined using the median operator. MSE: 0.1165. `

There isn't anything outstandingly irregular about the first time series, is there? Sorry, I'm new to this. Let me know if there's any more info I can offer to help, or if I'm using it wrong. Kevin

tschijvenaars commented 3 years ago

I had the same problem. I tried everything, but I think it is a faulty implementation of the mlp() and elm() functions when they use msts. When using a normal ts, I can specify 'start' as c(yyyy,dd), but when I tried the same for msts, I got this error. When I switched to year only (start=1996) I got no more error. Please look into this!