Step 2: Training Data

NeuroSolutions
NeuroSolutions v2.15
NeuroSolutions for Excel
Custom Solution Wizard
Interactive Book
Source Code License
Priority Support

GUI
NeuralBuilder
Step 1: Model Selection
Step 2: Training Data
Step 3: Cross Validation
Step 4: Topology Config.
Step 5: Layer Config.
Step 6: Simulation Control
Step 7: Probe Config.
Probing
Advanced Features
Developer Tools

This panel is used to select the training file(s) that contain the input and desired response data. The NeuralBuilder reads column-formatted ASCII files, such as Excel™ spreadsheets saved as text. Each column represents one channel of data and is tagged to be either an input, a desired response, a predictor, a symbol or skipped.

Tagging a column as Prediction uses the same data source for both the input and desired output. The difference is that the desired output is advanced by Delta time steps. For instance, suppose you have a year's worth of pricing data for a set of stocks. You may want to train the network by feeding each day's stock prices as the input and have the next day's stock prices (Delta=1) be the desired response.

Each column tagged as Symbol contains symbolic data instead of numeric data. The NeuralBuilder automatically encodes the symbols within the column into the 0's and 1's required for training.

Training Data Screen Shot