Transform data Rapfish. If you have successfully installed the software according to the Multidimensional Scaling Part 1 article, now I will provide an example of using the software. But keep in mind that, during the process of using it later, do not move any of the files from the extracted folder.
Before we start using the software, we need to prepare the data. Multidimensional Scaling uses ordinal and interval data. Ordinal data has a higher measurement level than nominal data and is considered qualitative data.
To make it easier to understand, MDS uses a score. Let’s say a score of 1–5, with 5 being the maximum or good score. Consequently, all data will be converted into these five ratings. It should be understood, however, that a score of 5 is not always good; in some cases, a score of 5 will be bad. For example, in the case of inflation or population growth rate, the current population growth rate is still considered bad. (We use the terms good, and bad only.)
What I will emphasize here is that the beginning of data preparation is the scoring technique, which determines the value of good or bad values. For example, inflation variable: 1–5, good value = 1, current value = 2.
Regional income variable: 1–5; good value = 5, current value = 3.
And so on…
Then, how is the technical method?
Take a look at my data below.
I have three districts that I want to rate as good or bad in the chili area suitability factor, for example. With variables 1 to 4. This is an example. Of course, it is not enough if it is only 4. Incidentally, the data collected is ratio data. Then I have to do the scoring first.
Var 1 and Var 2 have a good value of 5, meaning that the higher the value, the better. Conversely, var 3 and var 4 have a good value of 1. This means that the smaller the value, the better.
The first step I took was to determine the average, maximum, and minimum values. From these numbers, the class division value is obtained, namely: (Max – Min) / 4.
Why is there a number 4? Because of the min and max values, we will make the boundaries from class 1 to class 5 (there are 4 between classes).
Class limit 1 is the minimum value.
Class limit 2 is class limit 1 + class division.
Class 3 boundary is class 2 boundary + class division.
Class limit 4 is class limit 3 + class division.
Class limit 5 is class limit 4 plus class division. Which will be equal to the max value.
With that, we can determine that for var1.
Value 1: 0–6,109
Value 2: 6.110–10.302
Value 3: 10.303–14.495
Value 4: 14.496–18.688
Value 5: 18.689–22.881
We can determine for var 1 that District A has a value of 2, District B has a value of 1, and District C has a value of 5. If there are only 3 districts and 4 variables, we can still use our eyes to determine the scale value. If there are dozens of variables, of course your eyes will get sore.
You can use Excel using the if formula; you can also use SPSS. This time I will use SPSS because if I use Excel, the formula will be long. It’s better with SPSS because of the detailed stages that will be easier to understand.
Transform Data Rapfish: Prepare data rapfish
Let’s open the spss. We will convert ratio data into ordinal data.
The first step is to copy the initial data to the SPSS sheet.
click transform – recode into different variable
Select the variable to be transformed
Fill in the name and label, then click Change.
Click on old and new values, and then the class boundaries that we want to fill will appear. We select the range on the left side and enter the min and max values of the class in the value on the right side. Example:
Add all the class boundaries, then click continue.
It will return to the previous window; click OK. In the spss sheet, var1_rev has appeared instead of var1.
Do this to all variables (Only 4 as an exercise). Don’t forget the difference for variables 3 and 4 because the good is at value 1, then the scale is reversed, and the value of 5 that we have in variable 1 when transformed becomes value 1. Likewise for variable 4.
Example of variable 4
Final result
Using SPSS is easier, but the implementation is long if there are many variables. It is better to keep using the Excel formula by using “if”. It usually takes a long time at the beginning, but if you have succeeded, the formula can be copied and pasted so that it is faster.
I think this is it for part 2. In the next part, I will go directly to the Rapfish add-ins and multidimensional scaling software. So don’t forget to prepare the data first.
Continued: Multidimensional Scaling, Part 3
Leave a Reply