I am so thankful for the help that I've been able to get here so far. I'm hoping that someone can help me with another issue. I have these distribution curves that go out for 20 periods. The curves vary in type but I'm just starting to try to extend the standard bell curve first. I need it to extend out to at least 36 periods (my x axis values would be 0 to 36). What I need to do is determine the various y values on a scale of 0% to 100% while keeping the bell curve in tact. I've tried using averages to increase the number of values using the 18 period chart and obtaining the average of every 2 points to determine a point between them but realized this causes my curve to have a step pattern to it instead of a smooth slope. I guess I need to find a way to weight the average toward the higher value but I'm really struggling on how to accomplish my task. I've also been playing with the standard deviation but I can't seem to come up with anything that makes sense.

Does anyone know how to determine the points so I can have a smooth bell curve? I'm also going to have to perform this function on a front and rear loaded curves as well and have no idea how I'm going to approach those yet.

Thanks in advance.

Hi, I think that you should be talking to a statistician.

Excel has some ability to extend series.

For example if you have a standard line chart (graph), right-click on the series line and select 'Add Trendline ..."

Then you can select an algorithm that best fits your data.

Once the trendline is in place, right-click it and select 'Format Trendline' In the Options tab, you can then extend the trendline to future or~~paste~~past periods.Regarding a 'Bell curve', my understanding is that this was a plot of existing data, and shows whether the data has a 'Normal' or Gaussian distribution. It isn't something that can be extended. You are plotting the number of measurements that occur for any given value or range of values.

Adding more real data points will improve the accuracy of the plot, but adding calculated data isn't appropriate, unless the plot is being prepared for demonstration purposes. But again, I think that you need to talk to a statistician - which I am not.

Regards

Thanks for your help. After I posted I was thinking about that as well so I did some searching and may have come up with a good place to get some help but I'm not sure if what I want to do is going to work the way I need it to. I appreciate your help with this though. If you decide to become a statistician let me know! LOL!!

Thanks.

Ask Your Question

Weekly Poll

Do you think Samsung's Bixby will compete well against other phone AI systems?

Discuss in The Lounge

Poll History