Please take this survey to help us learn more about how you use third party tools. Your input is greatly appreciated!

Vertica ML GET_MODEL_SUMMARY got error

I am new to Vertica ML model, I have try LOGISTIC_REG model, i can create the model. but when I do
select GET_MODEL_SUMMARY(USING PARAMETERS model_name='iot_test');
got the error said:
ODBC Database Error: ERROR 3457: Function GET_MODEL_SUMMARY() does not exist, or permission is denied for GET_MODEL_SUMMARY()

Best Answers

  • Vertica_CurtisVertica_Curtis Employee
    Answer ✓

    Great! Don't hesitate to reach out again if you run into questions. Glad to help.


  • SruthiASruthiA Employee

    what is your vertica version?

  • V8

  • v8.1.1. thanks for quick response

  • SruthiASruthiA Employee

    GET_MODEL_SUMMARY was introduced in 9.0. Hence what you are seeing is expected as vertica 8.x doesn't contain that function.

  • thanks!

  • SELECT apn,is_iot, predict_logistic_reg(bytesup,bytesdown
    USING PARAMETERS model_name='iot_test_ds'
    AS Prediction
    FROM tmp_iot_train;
    got the error:
    ERROR 3457: Function predict_logistic_reg( int, int) does not exist, or permission is denied for predict_logistic_reg(int, int)

    any thought on this?

  • SruthiASruthiA Employee

    Are apn, is_iot columns from tmp_iot_train table? Can you share me the output of following

    SELECT SUMMARIZE_MODEL('iot_test_ds');

  • Hi, Sruthia: I sent the result to you via email last Friday, if you did not receive it please let me know


  • I found the issue: need to put "public" before procedure. it works now.. thanks all your great support!

  • Hi,

    now I tried to first balance data then normalize data, all these steps work fine.
    but when I tried to create the model and I got error: "Unsupported column type [Numeric(2,2)] for column [bytesup_home]"
    what is data type need for this model?

    SELECT BALANCE('balance_iot_balance_data', 'tmp_iot_wa_cls_train_data', 'is_iot', 'under_sampling'
    USING PARAMETERS sampling_ratio = 0.05);
    SELECT NORMALIZE('normalized_iot_balance_data', 'balance_iot_balance_data', 'bytesup_home,bytesup_work,bytesdown_home,bytesdown_work,dur_home,dur_work', 'minmax');
    -----------format column
    create or replace view normalized_iot_balance_data_v as
    select imei,firstcgi,apn,cfrc
    ,bytesup_home::NUMERIC(2, 2) bytesup_home
    ,bytesup_work::NUMERIC(2, 2) bytesup_work
    ,bytesdown_home::NUMERIC(2, 2) bytesdown_home
    ,bytesdown_work::NUMERIC(2, 2) bytesdown_work
    ,dur_home::NUMERIC(2, 2) dur_home
    ,dur_work::NUMERIC(2, 2) dur_work
    from normalized_iot_balance_data;

    USING PARAMETERS ntree=50, sampling_size=0.2
    DBC Database Error: ERROR 8135: Problem in rf_classifier.
    Detail: Cannot compute input column list.
    Detail: Unsupported column type [Numeric(2,2)] for column [bytesup_home]

  • SruthiASruthiA Employee

    What you are seeing is expect. We don't support NUMERIC type for RF_CLASSIFER in 8.1.1. We support it in 9.2. please check predictor_columns Argument Description in the below links

  • great, Sruthia.. thanks for your lighting response....

  • Thanks!

  • In V8, any function can be used for evaluating feature importance? like RF_PREDICTOR_IMPORTANCE in v9

  • In V9, has any function to identify similar variables and Removing redundant features so we can increase accuracy?

  • totally agree with you! thanks... based on my current study, RF already gave me pretty accurate predication and I just want to explore more possible feature in V8. we will upgrade to V9 soon, we will test all these new features.

    thanks again, all your feedback were very helpful!!

  • actually this is a first time I used Vertica support forum, you guys did awesome job! was much and much better than oracle support, lol
    thanks again and have a good weekend....

Leave a Comment

BoldItalicStrikethroughOrdered listUnordered list
Align leftAlign centerAlign rightToggle HTML viewToggle full pageToggle lights
Drop image/file

Can't find what you're looking for? Search the Vertica Documentation, Knowledge Base, or Blog for more information.