Announcing general availability of Native Scoring using PREDICT function in Azure SQL Database

Today we are announcing the general availability of the native PREDICT Transact-SQL function in Azure SQL Database. The PREDICT function allows you to perform scoring in real-time using certain RevoScaleR or revoscalepy models in a SQL query without invoking the R or Python runtime.

The PREDICT function support was added in SQL Server 2017. It is a table-valued function that takes a RevoScaleR or revoscalepy model & data (in the form of a table or view or query) as inputs and generates predictions based on the machine learning model. More details of the PREDICT function can be found here.

 /* Step 1: Setup schema */
drop table if exists iris_data, iris_models;

create table iris_data (
              id int not null identity primary key
              , "Sepal.Length" float not null, "Sepal.Width" float not null
              , "Petal.Length" float not null, "Petal.Width" float not null
              , "Species" varchar(100) null

create table iris_models (
       model_name varchar(30) not null primary key,
       model varbinary(max) not null,
       native_model varbinary(max) not null

/* Step 2: Populate test data from iris dataset in R */
insert into iris_data
("Sepal.Length", "Sepal.Width", "Petal.Length", "Petal.Width", "Species")
execute   sp_execute_external_script
                     @language = N'R'
                     , @script = N'iris_data <- iris;'
                     , @input_data_1 = N''
                     , @output_data_1_name = N'iris_data';

/* Step 3: Create procedure for training model */
create or alter procedure generate_iris_model
(@trained_model varbinary(max) OUTPUT, @native_trained_model varbinary(max) OUTPUT)
       execute sp_execute_external_script
         @language = N'R'
       , @script = N'
# Build decision tree model to predict species based on sepal/petal attributes
iris_model <- rxDTree(Species ~ Sepal.Length + Sepal.Width + Petal.Length + Petal.Width, data = iris_rx_data);

# Serialize model to binary format for storage in SQL Server
trained_model <- as.raw(serialize(iris_model, connection=NULL));

# Serialize model to native binary format for scoring using PREDICT function in SQL Server
native_trained_model <- rxSerializeModel(iris_model, realtimeScoringOnly = TRUE)
       , @input_data_1 = N'
select "Sepal.Length", "Sepal.Width", "Petal.Length", "Petal.Width", "Species"
from iris_data'
       , @input_data_1_name = N'iris_rx_data'
       , @params = N'
@trained_model varbinary(max) OUTPUT, @native_trained_model varbinary(max) OUTPUT'
       , @trained_model = @trained_model OUTPUT
       , @native_trained_model = @native_trained_model OUTPUT;

/* Step 3: Train & store a decision tree model that will predict species of flowers */
declare @model varbinary(max), @native_model varbinary(max);
exec generate_iris_model @model OUTPUT, @native_model OUTPUT;

delete from iris_models where model_name = 'iris.dtree';
insert into iris_models (model_name, model, native_model) values('iris.dtree', @model, @native_model);

select model_name
     , datalength(model)/1024. as model_size_kb
     , datalength(native_model)/1024. as native_model_size_kb
  from iris_models;

/* Step 4: Generate predictions using PREDICT function */
declare @native_model varbinary(max) =
        (select native_model from iris_models where model_name = 'iris.dtree');
select p.*, d.Species as "Species.Actual",
  from PREDICT(MODEL = @native_model, DATA = dbo.iris_data as d)
  with(setosa_Pred float, versicolor_Pred float, virginica_Pred float) as p;