Feature Scaling - Platform For AI - Alibaba Cloud Documentation Center

The Feature Scaling component can scale dense or sparse numeric data by using common scaling functions.

Overview

The Feature Scaling component has the following characteristics:

Supports common scaling functions such as log2, log10, In, abs, and sqrt.
Supports dense and sparse data.

Configure the component

You can use one of the following methods to configure the Feature Scaling component.

Method 1: Configure the component on the pipeline page

You can configure the parameters of the Feature Scaling component on the pipeline page of Machine Learning Designer of Machine Learning Platform for AI (PAI). Machine Learning Designer is formerly known as Machine Learning Studio. The following table describes the parameters.


Tab	Parameter	Description
Fields Setting	Scaled Features	The features that you want to scale.
	Label Column	The label column. If this parameter is specified, the x-y histogram that displays the relationship between the features and the objective variables can be viewed.
	Sparse Features (K:V,K:V)	Specifies whether the training data is sparse. If the data is sparse, a single field contains all the data instead of a single data record.
	Reserve Converted Features	Specifies whether to prefix new features with scale_.
Parameters Setting	Scaling Function	The Feature Scaling component supports the following scaling functions: log2 log10 ln abs sqrt

Method 2: Use PAI commands

Configure the component parameters by using PAI commands. You can use the SQL Script component to call PAI commands. For more information, see SQL Script.

PAI -name fe_scale_runner -project algo_public
    -Dlifecycle=28
    -DscaleMethod=log2
    -DscaleCols=nr_employed
    -DinputTable=pai_dense_10_1
    -DoutputTable=pai_temp_2262_20380_1;


Parameter	Required	Description	Default value
inputTable	Yes	The name of the input table.	None
inputTablePartitions	No	The partitions that are selected from the input table for training. Set this parameter in the `Partition_name=value` format. To specify multi-level partitions, set this parameter in the `name1=value1/name2=value2;` format. If you specify multiple partitions, separate them with commas (,).	All partitions in the input table
outputTable	Yes	The output table after scaling.	None
scaleCols	Yes	The features that you want to scale. Sparse features are automatically displayed. You can select only the features of numeric data types.	None
labelCol	No	The label column. If this parameter is specified, the x-y histogram that displays the relationship between the features and the objective variables can be viewed.	None
categoryCols	No	The selected fields that are processed as enumerated features. These fields do not support scaling.	""
scaleMethod	No	The method that is used for scaling. Value values: log2 log10 ln abs sqrt	log2
scaleTopN	No	If you do not set the scaleCols parameter, the system automatically selects the top N features that require scaling.	10
isSparse	No	Specifies whether features are sparse features in the key-value format.	Dense data
itemSpliter	No	The delimiter that is used to separate sparse key-value pairs.	,
kvSpliter	No	The delimiter that is used to separate sparse keys and values.	:
lifecycle	No	The lifecycle of the output table.	7
coreNum	No	The number of cores. The value of this parameter must be a positive integer. Valid values: [1,9999]. This parameter must be used together with the memSizePerCore parameter.	Determined by the system
memSizePerCore	No	The memory size of each core. Unit: MB. The value of this parameter must be a positive integer. Valid values: [2048,64 × 1024].	Determined by the system

Examples

Input data

Execute the following SQL statements to generate input data:

create table if not exists pai_dense_10_1 as
select
    nr_employed
from bank_data limit 10;

Parameter settings
On the Fields Setting tab, set the Scaled Features parameter to nr_employed. Only the features of the numeric data types are supported. On the Parameters Setting tab, set the Scaling Function parameter to log2, as shown in the following figure.
Results
nr_employed
12.352071021075528
12.34313018339218
12.285286613666395
12.316026916036957
12.309533196497519
12.352071021075528
12.316026916036957
12.316026916036957
12.309533196497519
12.316026916036957

nr_employed
12.352071021075528
12.34313018339218
12.285286613666395
12.316026916036957
12.309533196497519
12.352071021075528
12.316026916036957
12.316026916036957
12.309533196497519
12.316026916036957