A number of factors go into deciding a binning strategy. Having fewer values typically leads to a more compact model and one that builds faster, but it can also lead to some loss in accuracy.
Model quality can improve significantly with well-chosen bin boundaries. For example, an appropriate way to bin ages might be to separate them into groups of interest, such as children 0-13, teenagers 13-19, youth 19-24, working adults 24-35, and so on.
Table 4-4 lists the binning techniques provided by Oracle Data Mining.
Table 4-4 Binning Methods in DBMS_DATA_MINING_TRANSFORM
| Binning Method | Description |
|---|---|
|
Top-N Most Frequent Items |
You can use this technique to bin categorical attributes. You specify the number of bins. The value that occurs most frequently is labeled as the first bin, the value that appears with the next frequency is labeled as the second bin, and so on. All remaining values are in an additional bin. |
|
Supervised Binning |
Supervised binning is a form of intelligent binning, where bin boundaries are derived from important characteristics of the data. Supervised binning builds a single-predictor decision tree to find the interesting bin boundaries with respect to a target. It can be used for numerical or categorical attributes. |
|
Equi-Width Binning |
You can use equi-width binning for numerical attributes. The range of values is computed by subtracting the minimum value from the maximum value, then the range of values is divided into equal intervals. You can specify the number of bins or it can be calculated automatically. Equi-width binning should usually be used with outlier treatment. (See "Routines for Outlier Treatment".) |
|
Quantile Binning |
Quantile binning is a numerical binning technique. Quantiles are computed using the SQL analytic function |