Limitation in Excel, Will it be fixed in Office 12?

[alexaronson]

Well I was playing with some forecast data and I tried plotting 36,000 items vol vs. the MAPE statistic for dispersion. Excel gives me an error alert saying that I can only plot 32,000 items per times series. I know I can add another time series for the balance and change the color and shape so it matches the first time series.

Does anyone know if O12 will allow for more items to be graphed in a time series?

I mean, gosh, A million rows, I know someone will try graphing that many items on a single graph line or scatter plot.

 

[Jon Peltier]

Question: If you could in fact plot 36k points, could you see what was happening? There's not too many more pixels than that in a medium sized chart.

As far as I have heard, Excel 12 will still have the limit of 32k plotted points per series.

 

[alexaronson]

The forecast system that I am working to develope will be forecasting about 3.2 million forecast each month for 18 months out. So I need a sample size that will represent the population. The items that I am forecasting are not consistant, very lumpy, and if the system can generate a base forecast of about 40% accuracy, then the system will do what it needs to do.

 

[Greg Truby]

It's been a while since my last stats brush up, but what increase in your level of confidence level at a given interval or decrease in your confidence interval at a given confidence level are you going to achieve by increasing your sample size from 32K to 36K? I don't recall how much a severely skewed distribution would impact this, but at least my first reaction is to scratch my head and wonder how much you boost confidence levels moving n from 30 or 32K to 36K.

 

[alexaronson]

To be honest, I have no idea. I had the data available, so I decided to run with it. I do know that if I were to take 100 random samples of 30 data points, I would have a mean and standard deviation that would jump all over the place and I would not have any real confidence in what those samples were telling me.

Interesting to note, shortly after the original post I decided to ungroup my data from one single mass to individual segments (9 in all). The largest group had still made up 23,000 items. However, each of the segments developed a recognizable bell curve compared to the single mass scatter plot which was real ugly.