Linear regression Forecasting in R

Question

Having trouble scripting my forecast in R. I have all my time series data in a .csv document that is imported to the global environment. The code works all the way down to anova(reg1). So my question is really why the rest of the script doesn't work, and how do I need to edit the script to make it look something like example 4.4 looks like in this link(https://www.otexts.org/fpp/4/8)

data <- read.csv("Ny.csv", h=T, sep=";")

SeasonallyadjustedStockprice<-data$Seasonallyadj

Oilprice<-data$Oilaverage

Index<-data$DJI

plot.ts(Oilprice)

plot.ts(SeasonallyadjustedStockprice)

plot.ts(Index)

reg1<-lm(SeasonallyadjustedStockprice~Oilprice+Index)

summary(reg1)

anova(reg1)

Time <- tslm(SeasonallyadjustedStockprice~Oilprice+Index) 

f <- forecast(Time, h=5,level=c(80,95))

The error message is the following: Error in tslm(SeasonallyadjustedStockprice ~ Oilprice + Index) : Not time series data

So what exactly is your question here? Please edit your post to make that more clear. Make sure to supply a [reproducible example](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example) that doesn't rely on data only available on your computer. — MrFlick, Dec 28 '15 at 21:32
What do you mean by "doesn't work"? Are you getting an error message? If so, you should include that. — MrFlick, Dec 28 '15 at 21:54

score 4 · Accepted Answer · answered Dec 28 '15 at 22:10

It looks like its because the terms on the right side of the formula are not time series. Here is a simplified example:

>df
  a b
1 1 2
2 2 4
3 3 6
4 4 8

> y
[1] 1 3 5 7

>tslm(y ~ a + b, data=df)
Error in tslm(y ~ a + b, data = df) : Not time series data

> dfts = ts(df)    # This is of class 'mts' (multiple time series)
> tslm(y ~ a + b, data=dfts)

Call:
lm(formula = formula, data = dfts, na.action = na.exclude)

Coefficients:
(Intercept)            a            b  
         -1            2           NA

Which is basically sane output for this overly simplified example.

Linear regression Forecasting in R

1 Answers1