Error when trying to load data from SQL datatable into DataView

Question

I have created a SQL table with below code:

CREATE TABLE AAPL_1day
(
    [INDEX] INT IDENTITY(1,1) PRIMARY KEY,
    [DateTime] SMALLDATETIME NOT NULL,
    [MonthNr] FLOAT NOT NULL,
    [DayNr] FLOAT NOT NULL,
    [roc_1] FLOAT,
    [roc_2] FLOAT,
    [hypo_upordown7days] FLOAT
);

First rows in the table looks like this:

The problem I have now is to load this table into the variable "data" below.

I use var preview = data.Preview(); to be able to see if there is any data loaded into "data". However var preview is null.

I get the error below when running the code:

System.InvalidCastException: 'Specified cast is not valid.'

All columns in databaseName: AAPL have the datatype FLOAT. In the class "AAPLData" I have assigned "float" to all columns.

I can't figure out what I am doing wrong. I connect to the database and table successfully with the connection string and table name.

Even if I try to only retrieve one column, for example the below, I still get the same error:

[LoadColumn(2)] public float MonthNr;

Code:

void loadInData()
{
    // Set up MLContext
    var mlContext = new MLContext();

    // Define database connection parameters
    string databaseName = "AAPL";
    string connectionString = "Server=localhost;Database=" + databaseName + ";Trusted_Connection=True;";
    var tableName = "AAPL_1day";

    var loader = mlContext.Data.CreateDatabaseLoader<AAPLData>();
    var source = new DatabaseSource(SqlClientFactory.Instance, connectionString, $"SELECT * FROM {tableName}");
    var data = loader.Load(source);
    var preview = data.Preview(); //System.InvalidCastException: 'Specified cast is not valid.'

    string dummy_breakpoint = ""; //Break point here
}

public class AAPLData
{
    [LoadColumn(2)] public float MonthNr;
    [LoadColumn(3)] public float DayNr;
    [LoadColumn(4)] public float roc_1;
    [LoadColumn(5)] public float roc_2;
    [LoadColumn(6), ColumnName("Label")] public float hypo_upordown7days;
}

Answer 1

According to the following SO answer, a SQL float is actually a double, so the data type coming back is a double and you'd need to convert it to a float.

Why is a SQL float different from a C# float