Hello,

I need to get a prediction for 4 days ahead for an item price, I have only the last 10 days prices history of that item.

what would be the equation to solve this ?

Thanks

Printable View

- July 28th 2009, 02:53 AMisaacm1predict future price
Hello,

I need to get a prediction for 4 days ahead for an item price, I have only the last 10 days prices history of that item.

what would be the equation to solve this ?

Thanks - July 28th 2009, 03:52 AMpickslides
There are many different ways to predict the next 4 values. It can be dependant on many things, what were the previous 10 values you have?

- July 28th 2009, 05:03 AMisaacm1
11.4124 , 11.4196 , 11.4289 , 11.4127, 11.4076 , 11.3988 , 11.3821 ,11.3798 , 11.3926 ,11.4025

of course most accurate method will be better but please post two if possible

Thanks - July 28th 2009, 01:16 PMisaacm1
any reply ??

- July 28th 2009, 02:12 PMpickslides
Be patient my friend, they are many posters from all different corners of the world on this forum so everyone is on a different time zone, since my last post I have enjoyed some very deep sleep, 8 of the best you could say.

Looking at your data and considering it is daily I would suggest the next 4 values could be decreasing. There seems to be a peak at the 3rd value and then after the 8th value it increases again. 10 values is a very small set to make a prediction on. There seem to be 2 weekly increases after values 1,2,8 & 9. These could very well be weekends.

Any more information regarding the nature of the data? - July 28th 2009, 03:09 PMisaacm1
sorry, the numbers are prices, it is a challenge question from someone nothing more

any reasonable prediction will do

Thanks - July 28th 2009, 03:19 PMpickslides
- July 28th 2009, 04:17 PMisaacm1
thanks but I need equation , he also mention to use the average of all 10 days , does that help ?

- July 28th 2009, 05:54 PMpickslides
For your data

11.4124

11.4196

11.4289

11.4127

11.4076

11.3988

11.3821

11.3798

11.3926

11.4025

you could use a linear least squares regression, there are many options. This would give the equation - July 29th 2009, 12:29 AMisaacm1
Thanks, where x is the number of future days ?

- July 29th 2009, 01:51 PMpickslides
yep, x=11,12,13,14 for the next 4 values. This result is linear and will tend to zero as

- July 29th 2009, 02:06 PMisaacm1
Many Thanks,

one question please how you get the -0.0038 value ? - July 29th 2009, 02:25 PMpickslides
It is the gradient for y=bx+a from the least squares regression. It is a long a boring explantion but if you really want to know check out

Least-Squares Line - July 29th 2009, 02:50 PMisaacm1
if my numbers changed I need to know how to get the "b" value (-0.0038)

- July 29th 2009, 02:58 PMpickslides