Prev: regular expression
Next: structure of data uicontrol
From: Lucas Wall on 4 Aug 2010 09:25
Hello:

I am new to matlab and I would like to get some help on choosing a tool to predict events. I have a number of subjects for which I know which of four events occurred in the past. I have some indicators that believe help in explaining when those subjects go through any of the four events. I would like to know which tool to use to create a predicting model that would give me probabilities for each event. I tried mnrfit and the indicators are statistically sound. Can I use the betas to predict those events? Any other model would work better?

Thanks.

Example:

Event Observations
Subject | EventA | EventB | EventC | EventD
ABC | Yes | No | No | No
DEF | No | Yes | No | No
GHI | No | No | No | Yes

Indicators
Subject | Indicator1 | Indicator2 | Indicator3
ABC | 100 | 25 | 34
DEF | 21 | 110 | 80
GHI | 34 | 23 | 98

Which model can be used that can help predict probabilities of other subjects going through event A, B, C, or D?
From: Adam Chapman on 4 Aug 2010 11:02
On Aug 4, 2:25 pm, "Lucas Wall" <wallllu...(a)hotmail.com> wrote:
> Hello:
>
> I am new to matlab and I would like to get some help on choosing a tool to predict events.  I have a number of subjects for which I know which of four events occurred in the past.  I have some indicators that believe help in explaining when those subjects go through any of the four events.  I would like to know which tool to use to create a predicting model that would give me probabilities for each event. I tried mnrfit and the indicators are statistically sound.  Can I use the betas to predict those events?  Any other model would work better?
>
> Thanks.
>
> Example:
>
> Event Observations
> Subject | EventA | EventB | EventC | EventD
> ABC | Yes | No | No | No
> DEF | No | Yes | No | No
> GHI | No  | No | No | Yes
>
> Indicators
> Subject | Indicator1 | Indicator2 | Indicator3
> ABC | 100 | 25 | 34
> DEF | 21 | 110 | 80
> GHI | 34  | 23 | 98
>
> Which model can be used that can help predict probabilities of other subjects going through event A, B, C, or D?

If you have a large amount of example data, try using a neural net.
They're interesting.
 | 
Pages: 1
Prev: regular expression
Next: structure of data uicontrol