Modelling: Linear or Nonlinear Time Series

[morganstark]

Suppose I have a box contains 5 balls numbered from 1 to 5, and the ball is randomly drawn with replacement. Everyday 10 drawings are made, forming a time series sequence like

1, 5, 1, 4, 3, 2, 4, 2, 5, 5

I was given 2000 historical drawings for the last 200 days, and I'm free to use any amount of them to forecast the next drawing.

I would like to know whether i should model it as a linear time series, or a nonlinear time series? Thank you.

 

[JeffM]

Your question is quite unclear. What is the purpose of the model? If the underlying process is random enough, the line of best fit will be a horizontal line with equation

[MATH]x \approx 3 + 0 * t.[/MATH]
The whole purpose of a time series is to show that a process is not consistent with respect to time, but your hypothesized situation is consistent with respect to time. What do you expect to demonstrate with a time series?

 

[tkhunny]

Suppose I have a box contains 5 balls numbered from 1 to 5, and the ball is randomly drawn with replacement. Everyday 10 drawings are made, forming a time series sequence like

1, 5, 1, 4, 3, 2, 4, 2, 5, 5

I was given 2000 historical drawings for the last 200 days, and I'm free to use any amount of them to forecast the next drawing.

I would like to know whether i should model it as a linear time series, or a nonlinear time series? Thank you.

A linear time series is quite inappropriate, just like any other model. Successive draws are INDEPENDENT. No amount of prior information has ANYTHING to do with the next draw. The process is MEMORYLESS.

This is not to say that suckers have not paid lots of money to buy into such fraudulent schemes. I would personally recommend neither inventing nor participating in such pursuits.

 

[morganstark]

A linear time series is quite inappropriate, just like any other model. Successive draws are INDEPENDENT. No amount of prior information has ANYTHING to do with the next draw. The process is MEMORYLESS.

This is not to say that suckers have not paid lots of money to buy into such fraudulent schemes. I would personally recommend neither inventing nor participating in such pursuits.

If i believe there's a fraudulent schemes, and my EQ/IQ is not TOO LOW, then I'll make a report instead of step in front and reply like a sucker.

 

[tkhunny]

Try to read what is actually written. I pointed an accusing finger in no specific direction. If you want to be offended by the facts, that's up to you.

If you believe there is something of interest in Lottery predictor algorithms, you do not understand the Lottery or the Lottery is cheating in favor of someone.

I said nothing of who the sucker might be - only that there are some. Should I have used "victim", instead? If you are the one sponsoring the scheme, then you would be the perpetrator and your customers would be the victims.