Hi,

Just wondering if anyone has used SPSS (statistical program) in developing a a rating system.

I think it is possible to use this program to determine how much each variable contributes to the overall success of a horse.

I would be interested to speak to anyone that has tried to create a model using stastistical software like this rather than selecting a number of variables and running it through a database to get the results.

Any feedback would be greatly appreciated.

Blocka39

Just had a quick chat to a friend of mine (who uses SPSS at work) about the contents of your email blocka39. She started talking double-dutch & using words I couldn't comprehend (there's many of those!). In a nutshell the answer was 'yes' but I think you need to be a bit of a rocket scientist to use. Maybe one of our forumites can suggest something a bit easier to use.

__________________
Ta me go maith

Hi Blocka39,

I've had quite a bit of experience with SPSS, and I'm sure it could be of benefit. (For those who haven't heard of it, SPSS measures and reports on the validity of relationships between variables).

The biggest problem that I see is how those variables are imput into the programme. The data must be in numeric form only -- not a problem for weight, TAB number, finishing position, etc. But for variables such as class, track condition, jockey, trainer and venue, the data must be converted into numbers.

Another problem would be that the data you are analysing is of different types, even though it is all numeric. For example, $4 is worth twice as much as $2. A 2400m race is twice as long as a 1200m race. But if you allocate a fast track a "1" and a heavy track a "5", then a slow track ("4") is not worth twice as much as a good track ("2"). While SPSS can handle such different data types (specifically nominal, ordinal and interval data) the way that you code the data will affect the outcome of the tests. Interpretation will be difficult.

That said, it certainly has possibilities. You need a racing fan who is a stats expert to combine the two. But best of luck to you for looking "outside the square".

Monkeyinjapan

Thanks for the replies.

I have previously used SPSS as an engineering application for regression modelling and thought there could be some way to adapt it to racing.

I understand that there are a number of hurdles associated with how the data can be entered. I have come up with a few ideas but i havent thought about it to thoroughly yet. In order for it to work i think it is going to involve a fair bit of data manipulation.

Blocka39

Hi,
In another thread I mentioned the Indexing of Data, some packages require this to be done ,as has already been mentioned here. Some of the benefits other than being able to input are as follows,
Indexing allows you to place data on a comparable scale
It will allow you to determine the relative importance of each Index as a predictor.
Allows you to combine Indexes on a weighted basis to create an Overall Ability Indicator (using ROE weights)
The steps in creating an Index are as follows; you need to determine which way is up? with this data is the HIGHER the number is, the better it is, and it should therefore have a Higher Index number.
For each Factor you will need to either find the Maximum number or Minimum number benchmark that the other data will be compared to or use the Max an Min. within the data.Now depending on "which way is up"
Using Time as an example so the smaller the number the higher the score

Using 3 runners runner a) 34.06 slowest runner and Max, runner b) 31.06, runner c)30.06 min data.
We will find the Index score for runner B)
Step 1 subtract runner b from Max value 34.06 -31.06 = 3sec
Step 2 divide step 1 result by the range of data (34.06 - 30.06=4 sec)
3sec divided by 4 sec =.75
Step 3 Multiply result of Step 2 by 40 (our index will have a min score of 100-60) .75 Times 40 = 30
Step 4 Add the result of Step 3 to 60 (our index will have a min score of 100 -40) , 30 plus 60 =90
so the the above runners A) will score 60 B) 90 and C) 100.
Just reverse the base numbers depending on "which way is up".
Hope this helps you organise your data