Solved – Error in optim – Auto.arima R

arimaforecastingrtime series

I am new to R and forecasting .I have data for a certain product. It contains value sales and its promotions. The data is weekly and there are about 104 data points.

I converted the sales into a ts object and created seasonaldummy's to capture seasonality. 

actual_val = ts(sku1$Sal , frequency = 52)
dummy_val = seasonaldummy(actual_val)

The dummy's were later combined with the promo variables to create external regressors xreg_val for the model. Those promotions which were not held for this sku were removed before combining the two.

model_value <- try(auto.arima(actual_val , xreg = xreg_val ) , silent = TRUE)

I have received the following error

Error in optim(init[mask], armaCSS, method = optim.method, hessian = FALSE,    
non-finite value supplied by optim

I could not understand where exactly I have gone wrong in this.

Attaching a sample of the data. Kindly help me with this

https://drive.google.com/file/d/0B6sOv1da0JMeb01XYW92UzRSZ0U/view?usp=sharing

Best Answer

The following works in the sense that it returns a valid model:

library(forecast)
sku1 <- read.csv("Sample.csv")
actual_val = ts(sku1$Sal , frequency = 52)
dummy_val = seasonaldummy(actual_val) 
model_value <- auto.arima(actual_val , xreg = dummy_val )

Perhaps your problem is that you haven't defined xreg_val.

However, it isn't a sensible model as you use 51 degrees of freedom for seasonality but have only two years of daily data. I suggest you use Fourier terms instead:

seas <- fourier(actual_val, K=10)
model_value <- auto.arima(actual_val , xreg = seas , lambda=0) 
plot(forecast(model_value, xreg=fourierf(actual_val, h=52, K=10), lambda=0))
Related Question