Calculating the average

[flbiker450]

Hello All,

I am new to your forum and I am not sure if this question is posted in thecorrect one. Hopefully if I am in thewrong one, it can be moved or I can be directed to the correct one.

My issue is this: I have a excel spreadsheet that is setup so when a question is asked it has to be weighted by 7 criteria’s. Each criterion is a separate column and ineach cell of that column is a numeric value. The first three columns have a numeric value from 1-4 and the last 4 havea numeric value from 1-3.
When the question is scored and the sum of each score isgenerated, it is check against a chart where a 1 = none, 2-5 = low, 6-8 =issue, 9-12 = risk. I believe it worksvery well and accurately projects the results of the question.
My question is this: If I add one more column (for a total of 8) to be used as a weighted scorewith the numeric value from 1-3, how much will that throw off my results?

Will it be much harder to achieve an issue or riskscore?
Will it through off all my metrics and score them to highor to low?
My boss feels it will not matter where I am pretty sureit will. I was wondering if there is provenmathematical theorems that would help me understand.

Any comments would be greatly appreciated.

Sorry I forgot t upload the workbook showing what I have.

Thank you
Dave

 

[stapel]

Since we don't know the parameters of your inputs, your rules or ratings, or your outputs, it is pretty much impossible to guess how much another input might "throw off" new output. Also, standard web-safety protocols would prevent sensible persons from opening an unknown attachments.

Kindly please reply with a clear statement of the exercise, the full instructions, and any required information (such as tables of values, ratings rules, graphics, etc). When you reply, please include a clear listing of your work and reasoning so far. Thank you!

 

[flbiker450]

??

I am sorry but did you open the spreadsheet? It is there so someone would be able to use it and work with it to see how it is working. There is no guessing, its straight forward.

Also, the rules or ratings are clearly defined "where a1 = none, 2-5 = low, 6-8 = issue, 9-12 = risk". If you open the spreadsheet an use it you will see how it works. If you read my comments while using the spreadsheet it will make sense. I uploaded the spreadsheet as a visual because any description alone would not work.

 

[Ishuda]

I am sorry but did you open the spreadsheet? It is there so someone would be able to use it and work with it to see how it is working. There is no guessing, its straight forward.

Also, the rules or ratings are clearly defined "where a1 = none, 2-5 = low, 6-8 = issue, 9-12 = risk". If you open the spreadsheet an use it you will see how it works. If you read my comments while using the spreadsheet it will make sense. I uploaded the spreadsheet as a visual because any description alone would not work.

As stapel mentioned, it is a bad idea to open any file unless you know the sender and, since I, in particular, don't know you, I would be loath to open it. But just for the fun of it, assume I would and that your equations and what you had written in the spread sheet made sense. That still wouldn't help unless we knew what kind of data base you were using the formulas and classifications on. But suppose we did know that. It still wouldn't help unless we knew what new method(s) and classification you were going to be adding, how it was classified, how it interacted with the other classifications, etc. In other words, you would have to write a manual for your work, including a description of the new classification you would use and how it would be used, make it understandable to other people outside the field you are working in, and provide a data set for testing (or, in mathematical terms, the domain and range of the classification functions, how they are related, and how they are used).

What I would suggest you do is run through a lot of cases you have done before using the present (old) method and hopefully can compare the results to what they should have been. For example, suppose you were trying to predict rain. If a test case said it would (would not) rain and it did (did not) rain, then it was a 'good result', otherwise it was a 'bad result'. Now run through the cases again with the new classification and method added and see how/if the results change. You could just keep a copy of future results using both methods for a while and compare results if you don't have a past database to use. Actually, it would seem to me that this had already been done before you changed an established procedure.