組件配置

您可以使用以下任意一種方式，配置邏輯回歸二分類組件參數(shù)。

方式一：可視化方式

方式二：PAI命令方式

PAI -name logisticregression_binary
    -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"

示例

1. 使用SQL語句，生成訓(xùn)練數(shù)據(jù)。

   drop table if exists lr_test_input;
   create table 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(1 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
       union all
           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
   ) a;

   生成的訓(xùn)練數(shù)據(jù)表lr_test_input如下。

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

2. 使用PAI命令，提交邏輯回歸二分類組件的訓(xùn)練參數(shù)。

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

3. 使用PAI命令，提交預(yù)測組件參數(shù)。更多參數(shù)說明，請參見預(yù)測。

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

4. 查看預(yù)測結(jié)果表lr_test_prediction_result。

   label	prediction_result	prediction_score	prediction_detail
   0	0	0.9999998793434426	{“0”: 0.9999998793434426, “1”: 1.206565574533681e-07}
   1	1	0.999999799574135	{“0”: 2.004258650156743e-07, “1”: 0.999999799574135}
   1	1	0.999999799574135	{“0”: 2.004258650156743e-07, “1”: 0.999999799574135}
   0	0	0.9999998793434426	{“0”: 0.9999998793434426, “1”: 1.206565574533681e-07}
   0	0	0.9999998793434426	{“0”: 0.9999998793434426, “1”: 1.206565574533681e-07}
   0	0	0.9999998793434426	{“0”: 0.9999998793434426, “1”: 1.206565574533681e-07}