Platform For AI:Logistic Regression for Multiclass Classification

Configure the component

You can use one of the following methods to configure the Logistic Regression for Multiclass Classification component.

Method 1: Configure the component on the pipeline page

Method 2: Use PAI commands

PAI -name logisticregression_multi
    -project algo_public
    -DmodelName="xlab_m_logistic_regression_6096"
    -DregularizedLevel="1"
    -DmaxIter="100"
    -DregularizedType="l1"
    -Depsilon="0.000001"
    -DlabelColName="y"
    -DfeatureColNames="pdays,emp_var_rate"
    -DgoodValue="1"
    -DinputTableName="bank_data"

Example

1. Execute the following SQL statements to generate training data:

   drop table if exists multi_lr_test_input;
   create table multi_lr_test_input
   as
   select
       *
   from
   (
       select
           cast(1 as double) as f0,
           cast(0 as double) as f1,
           cast(0 as double) as f2,
           cast(0 as double) as f3,
           cast(0 as bigint) as label
       union all
           select
               cast(0 as double) as f0,
               cast(1 as double) as f1,
               cast(0 as double) as f2,
               cast(0 as double) as f3,
               cast(0 as bigint) as label
       union all
           select
               cast(0 as double) as f0,
               cast(0 as double) as f1,
               cast(1 as double) as f2,
               cast(0 as double) as f3,
               cast(2 as bigint) as label
       union all
           select
               cast(0 as double) as f0,
               cast(0 as double) as f1,
               cast(0 as double) as f2,
               cast(1 as double) as f3,
               cast(1 as bigint) as label
   ) a;

   The following table provides the generated training data in the multi_lr_test_input table.

   f0	f1	f2	f3	label
   1.0	0.0	0.0	0.0	0
   0.0	0.0	1.0	0.0	2
   0.0	0.0	0.0	1.0	1
   0.0	1.0	0.0	0.0	0

2. Run the following PAI command to submit the parameters of the Logistic Regression for Multiclass Classification component:

   drop offlinemodel if exists multi_lr_test_model;
   PAI -name logisticregression_multi
       -project algo_public
       -DmodelName="multi_lr_test_model"
       -DitemDelimiter=","
       -DregularizedLevel="1"
       -DmaxIter="100"
       -DregularizedType="None"
       -Depsilon="0.000001"
       -DkvDelimiter=":"
       -DlabelColName="label"
       -DfeatureColNames="f0,f1,f2,f3"
       -DenableSparse="false"
       -DinputTableName="multi_lr_test_input";

3. Run the following PAI command to submit the parameters of the Prediction component:

   drop table if exists multi_lr_test_prediction_result;
   PAI -name prediction
       -project algo_public
       -DdetailColName="prediction_detail"
       -DmodelName="multi_lr_test_model"
       -DitemDelimiter=","
       -DresultColName="prediction_result"
       -Dlifecycle="28"
       -DoutputTableName="multi_lr_test_prediction_result"
       -DscoreColName="prediction_score"
       -DkvDelimiter=":"
       -DinputTableName="multi_lr_test_input"
       -DenableSparse="false"
       -DappendColNames="label";

4. View the multi_lr_test_prediction_result table.

   label	prediction_result	prediction_score	prediction_detail
   0	0	0.9999997274902165	{"0": 0.9999997274902165, "1": 2.324679066261573e-07, "2": 2.324679066261569e-07}
   0	0	0.9999997274902165	{"0": 0.9999997274902165, "1": 2.324679066261573e-07, "2": 2.324679066261569e-07}
   2	2	0.9999999155958832	{"0": 2.018833979850994e-07, "1": 2.324679066261573e-07, "2": 0.9999999155958832}
   1	1	0.9999999155958832	{"0": 2.018833979850994e-07, "1": 0.9999999155958832, "2": 2.324679066261569e-07}