Automatic Data Preparation

Most algorithms require some form of data transformation. During the model build process, Oracle Machine Learning for SQL can automatically perform the transformations required by the algorithm.

You can choose to supplement the automatic transformations with additional transformations of your own, or you can choose to manage all the transformations yourself.

In calculating automatic transformations, OML4SQL uses heuristics that address the common requirements of a given algorithm. This process results in reasonable model quality in most cases.

Binning and normalization are transformations that are commonly needed by machine learning algorithms.

Related Topics

Oracle Database PL/SQL Packages and Types Reference

Binning

Binning, also called discretization, is a technique for reducing the cardinality of continuous and discrete data. Binning groups related values together in bins to reduce the number of distinct values.

Binning can improve resource utilization and model build response time dramatically without significant loss in model quality. Binning can improve model quality by strengthening the relationship between attributes.

Supervised binning is a form of intelligent binning in which important characteristics of the data are used to determine the bin boundaries. In supervised binning, the bin boundaries are identified by a single-predictor decision tree that takes into account the joint distribution with the target. Supervised binning can be used for both numerical and categorical attributes.

Normalization

Learn about normalization.

Normalization is the most common technique for reducing the range of numerical data. Most normalization methods map the range of a single variable to another range (often 0,1).

How ADP Transforms the Data

The following table shows how ADP prepares the data for each algorithm.

Table 35-2 Oracle Machine Learning Algorithms With ADP

Algorithm	Machine Learning Function	Treatment by ADP
Apriori	Association rules	ADP has no effect on association rules.
CUR Matrix Decomposition	Feature selection	ADP has no effect on CUR Matrix Decomposition
Decision Tree	Classification	ADP has no effect on Decision Tree. Data preparation is handled by the algorithm.
Expectation Maximization	Clustering	Single-column (not nested) numerical columns that are modeled with Gaussian distributions are normalized. ADP has no effect on the other types of columns.
GLM	Classification and regression	Numerical attributes are normalized.
k-Means	Clustering	Numerical attributes are normalized.
MDL	Attribute importance	All attributes are binned with supervised binning.
MSET-SPRT	Classification (for anomaly detection)	Z-score normalization is performed.
Naive Bayes	Classification	All attributes are binned with supervised binning.
Neural Network	Classification and regression	Numerical attributes are normalized.
NMF	Feature extraction	Numerical attributes are normalized.
O-Cluster	Clustering	Numerical attributes are binned with a specialized form of equi-width binning, which computes the number of bins per attribute automatically. Numerical columns with all nulls or a single value are removed.
Random Forest	Classification	ADP has no effect on Random Forest. Data preparation is handled by the algorithm.
SVD	Feature extraction	Numeric attributes are centered if PCA is selected.
SVM	Classification, anomaly detection, and regression	Numerical attributes are normalized.
XG Boost	Classification and regression	ADP has no effect on XG Boost.