The NeuralNet function has a required input table and an optional weight table.
NeuralNetwork Input Table Schema
| Column Name |
Data Type |
Description |
|
input_column
|
Any numeric type |
Predictor variable. The table has one such column for each input column specified by the InputColumns argument. |
|
response_column
|
Any numeric type |
Response variable. The table has one such column for each input column specified by the ResponseColumns argument. |
NeuralNetwork Weight Table Schema
| Column Name |
Data Type |
Description |
|
group_by_column
|
Any numeric type |
Columns that contains neural network output for a group. The table has one such column for each input column specified by the GroupByColumns argument. |
| weight_n
|
DOUBLE PRECISION |
Initial weight for attribute n. The table has n such columns. |