Time series in small data with R

#1
Hi there
I have this data
year number of surgeries
2000 570
2005 1340
2006 4730
2007 5359
2009 6060
2010 6340
2014 ????
2015 ????
I want predict number of surgeries in 2014 and 2015.
I want use neural network, SVM and all methods for predicting data in R and then chose the best according to expert opinion.
I made prediction with Y=1046.9(LN(X))+ 4756, I removed year 2000 and 2005 as a Outlier and then fit this model Y=1046.9(LN(X))+ 4756
y=years X=number of surgeries
but I need more models for chossing the best. I need codes in R
Thanks in advance
Leila
 

noetsi

Fortran must die
#4
5 data points for prediction, where you also have a possible a trend - I would say this is hopeless.

regards
I did not see that, but this is certainly true for time series. You want at least 50 points as a bare minimum according to the common wisdom. Seasonality would be a major issue.
 
#5
I did not see that, but this is certainly true for time series. You want at least 50 points as a bare minimum according to the common wisdom. Seasonality would be a major issue.
Don't trust such rules of thumb.

For example my length has been considerably constant the last five years. And I believe that it make a good forecast for the coming years.

In other cases 10 000 observations is not enough, because of the inherent un-stableness of the series.