Structure and Types of Data
Data to be used in StatService is structured as a matrix, where columnss represent variables (the different algorithms, techniques or treatmets you want to compare), and rows represent cases specific situations or applications of the techniques. Each cell in this matrix represent the specific value used for comparison (e.g. execution time, quality of results, accuracy, etc.) for the variable and case in with it is located.
In STATServices, a numeral type of variables is assumed, i.e. by default data specified in cells is interpreted as floating point numbers.
In STATServices, you provide the data for analyses either as plain text or as a binary file, using formats such as Excel 2003/2007/2010. When using plain text input, thel formatting of the data can be parametrized by:
- Specifying the concrete separator token between cell items.
- Specifying the concrete separator token between rows.
- Specifying if you are going to use variable and a case names as the firts row and column of data.
|Variable 1||Cell Sep.||Variable 2||Cell Sep.||...||Cell Sep.||Variable N||Row Sep.|
|Case A||Data A-1||Cell Sep.||Data A-2||Cell Sep.||...||Cell Sep.||Data A-N||Row Sep.|
|Case B||Data B-1||Cell Sep.||Data B-2||Cell Sep.||...||Cell Sep.||Data B-N||Row Sep.|
|...||...||Cell Sep.||...||Cell Sep.||...||Cell Sep.||...||Row Sep.|
|Case X||Data X-1||Cell Sep.||Data X-2||Cell Sep.||...||Cell Sep.||Data X-N||Row Sep.|
Input Data creation examples
The image below shows a case in which we have 3 different sorting algorithms under, similar characteristics that will be under 3 differents scenarios. Algorithms are: Shell Sort, QuickSort and Distribution Counting.
In this case we used the space key to separate columns and CR (Carriage return) to separate rows.
Now we only have to select our CSV file and click on import:
The next image represents the result that might be shown on the web page for these last cases: