I have the data for Men's 400 meters winning times in Olympics for all years since 1896, and I'm supposed to predict the next winning time. However, the data is not linear, so what can I do to get the answer? I know linear regression would definitely not work. Also, the years from 1896 to 1960 seem so far off from the more recent years, that they are probably useless, but even the data from 1960 to 2008 doesn't form a perfectly uniform line, it's full of outliers. I know I didn't post the actual data here, but do you guys have any idea how I can approach this problem?
Also, I'm supposed to do this using the program called "R" so if you can give me tips with that, it would be great.

Thanks.