Why Recode?

"Recode" means literally "change or rearrange the code." Recoding can be useful in many situations in data mining projects. Here are some examples:

Missing values treatment: Missing values may be indicated by something other than NULL, such as "0000" or "9999" or "NA" or some other character string. One way to treat the missing value is to recode, for example, "0000" to NULL. Then Oracle Data Miner and the database recognize the value as missing.
Note: Do not use recode to treat missing values if the values are continuous; recode is not designed to deal with very large numbers of distinct values.
Change datatype (and value) of an attribute: For example, change "Y" or "Yes" to 1 and "N" or "No" to 0.
Establish a cutoff value: For example, recode all incomes less than $20,000 to the same value.
Group items: For example, group individual US states into regions. The "New England" region might consist of ME, VT, NH, MA, CT, and RI; to implement this, recode the five states to, say, NE (for New England).
Decrease the number of values of an attribute: For example, suppose the attribute LENGTH_OF_RESIDENCE has a numerical range from 1 to 34. In order to make the model build operation more efficient, you decide to consider only two classes of residence: "LOW" for residences of less than or equal to 10 years, and "HIGH" for residences of more than 10 years.

Recode Limitations

The following limitations apply to the recode transformation:

The Recode Wizard supports the relational operators =, <=, and >= only. This makes it difficult to partition values at a particular numerical value, for example: Income <= 100000 and Income > 100000). Recode is really meant for use with character strings and integers; use discretization to partition real numbers.
Recode works best with attributes of fairly low cardinality, that is, attributes with a relatively small number of distinct values. Oracle Data Miner does not allow you to define recode schemes for attributes with with more than 2000 distinct values. For attributes with more than 2000 distinct values, use discretization or the missing values transformation.