Randomize Integer

Goal

Write randomize rule logic to

• Takes the original values of the type integer, and adds random noise
• Specify the random noise by using the offset, flat-noise, and percentage-noise parameters

Ano Structure and Example

Structure

        ...
        randomize <column> <rule-name?>
            type integer
            format %d
            convert String2Integer
            offset? <offset?>
            flat-noise? <flat-noise?>
            percentage-noise? <percentage-noise?>
        ...

Update Example

table ROOMCATEGORY
    ...
    column integer MAXDISCOUNT size 10
    ...
...
task ... {
    update ROOMCATEGORY
        randomize MAXDISCOUNT
            type integer
            format %d
            convert String2Integer
            offset 5.0                  // add 5 to all original values
            flat-noise 0.0              // no flat noise
            percentage-noise 20.0       // adds 20% of the original value as size to the random noise
        ...
}

• type: integer
• format: %d - integer format
• convert: String2Integer - a built-in conversion function.
• offset: A fixed (double) value that is added to the original value.
• flat-noise: A fixed (double) noise value that is multiplied with the random number r
  • large flat-noise - larger spread of generated values
• percentage-noise: A percentage (0.0% - 100.0%) of the original value that is multplied with the random number r
  • large percentage-noise - larger spread of generated values

Noise Formula

1. A random number r is drawn (at runtime) from the Standard Normal (Gaussian) Distribution Function
2. We use r, together with given noise parameters to add noise to the original value

Noise formula for integer

original_value + offset + (flat-noise * r) + (original_value * percentage-noise/ 100) * r

The resulting value is cast from double to integer.