Skip to main content
cancel
Showing results for 
Search instead for 
Did you mean: 

Earn the coveted Fabric Analytics Engineer certification. 100% off your exam for a limited time only!

Reply
141507
Helper I
Helper I

ARIMA forecasting - Avoid negative preditions

Hi,

 

I am using below R code to generate Arima Forecast. The plot is successful and I dont want to see negative values in the plot. I used the lambda=0 in the arima() and my plot date is greater than zero records only. Still I am seeing negative predicions.

 

 

.forecast_R_Arima.JPG

 

Is there anyway to prevent this. Below is my R code.

 

 

 

library(xts)
library("forecast")
dates=as.Date(dataset$ReceivedDate,"%Y-%m-%d")
xs=xts(dataset$Volume,dates,frequency=12)
plot(xs)

arima1 <-arima(xs,order=c(2,1,5),seasonal=list(order=c(0,1,1),period=12,lambda=0))
q=forecast(arima1,h=60,level=c(99.5))
plot(q,col ="darkcyan",xaxt = 'n',xlab='Year', ylab = 'Volume',main='Inventory Forecast (Arima)')
axis(side=1,at=c(0,50,100,150,200),labels=c("Aug 2016","Oct 2016","Dec 2016","Jan 2017","Mar 2017"))
 
file.remove("D:/fitarima.csv")
write.table(q,file="D:/fitarima.csv", append=TRUE,sep=",",col.names=TRUE,row.names=TRUE)

Regards,

Senthil D

3 REPLIES 3
v-ljerr-msft
Employee
Employee

Hi @141507,

 

Here is an article about positive forecasts and forecasts constrained to an interval. Could you go to check if it helps in your scenario? Smiley Happy

 

Regards

Hi,

 

Thanks for your input, But using this, prediction line shows as constant.

 

Do We have any other way to achieve the positive prediction since we have positive values as resource data ?

 

Regards,

Senthil D

Hi,

 

Is any update for the above.

 

Regards,

Senthil D

Helpful resources

Announcements
April AMA free

Microsoft Fabric AMA Livestream

Join us Tuesday, April 09, 9:00 – 10:00 AM PST for a live, expert-led Q&A session on all things Microsoft Fabric!

March Fabric Community Update

Fabric Community Update - March 2024

Find out what's new and trending in the Fabric Community.