The chi^2 in my digram box (lower right hand corner) is 472 for 3 degrees of freedom. That's as statistically meaningful as they come. If you see anything wrong with what I've done (the dates the data come from are at the top of the chart) by all means, let's hear it.
Time Series Analysis
- Thread starter imtrader
- Start date
The chi^2 in my digram box (lower right hand corner) is 472 for 3 degrees of freedom. That's as statistically meaningful as they come. If you see anything wrong with what I've done (the dates the data come from are at the top of the chart) by all means, let's hear it.
The skew and kurtosis moves over time.
Hello guys,
I had originally asked the question regarding the timeseries. Now what I did to determine the independence of the series was perform auto correlation with different lags (from 1 to 20) and after doing that run the regression analysis to determine the time equation and nature of dependencies.
Unfortunately, I cant post the data. All I can say that the size of the two series is small (only 122 points).
I am wondering whether what I did is the correct thing?
Looking forward to hearing from you guys,
Thanks,
P.S - Petadollar, thank you for keeping the discussion active.
I had originally asked the question regarding the timeseries. Now what I did to determine the independence of the series was perform auto correlation with different lags (from 1 to 20) and after doing that run the regression analysis to determine the time equation and nature of dependencies.
Unfortunately, I cant post the data. All I can say that the size of the two series is small (only 122 points).
I am wondering whether what I did is the correct thing?
Looking forward to hearing from you guys,
Thanks,
P.S - Petadollar, thank you for keeping the discussion active.
You should start with something more general than autocorrelation, as absence of autocorrelation does not imply an independent time series. Sherry talks about this in his book. Also, if you have not already, compare your correlation coefficients to those you get with random, independent time series (repeatedly). This lets you see when your coefficients indicate dependence rather than statistical noise.