연구의 선진화와 생산성 향상에
필요한 최적의 소프트웨어와 컨설팅을
공급하도록 노력하겠습니다.

NeuroSolutions for Excel

NeuroSolutions for Excel is a Microsoft Excel add-in that simplifies and enhances the process of getting data into and out of a NeuroSolutions neural network. This tool benefits both the novice and the advanced neural network developer by allowing them to do everything from preprocessing the data, to designing the neural network, to generating reports of the results -- all directly from Excel!
The free evaluation edition of NeuroSolutions includes a demo of NeuroSolutions for Excel that will walk you through several examples. In addition, you can experiment with all of the features of this product using a sample of your own data. See how easy it is to tag your data, create and train a neural network and view a graphical report of the network's performance.

Introduction

NeuroSolutions for Excel is an add-in that enables you to develop your neural network models without ever having to leave the Excel environment. Everything you need is all contained within the NeuroSolutions sub-menu of Excel. In addition to the convenience factor, it offers functionality not available within the base NeuroSolutions software.
This tour will walk you through the process of solving a problem with NeuroSolutions for Excel. It will only scratch the surface of the features available, so you will want to download a free evaluation copy afterwards to get a more in-depth look.

Preprocess the Data

The first step in many problems is to preprocess the data to put it in a form that is most useful to the neural network. There are many different preprocessing functions that NeuroSolutions for Excel can perform, but here are three of the most commonly used ones:
  • Randomize Rows
    In many cases your data may be ordered such that similar inputs may be grouped together. This can cause a problem when training a neural network in that the network may be trained on one group of input data and then tested on a group of data that has little similarity to the training data. You want to try to have your training data be representative of the entire data set. NeuroSolutions for Excel can handle this by randomizing the rows of the data set before training.
  • Translate Symbolic Columns
    Neural networks require that you feed them numeric input and desired output data. There are many problems where you want the neural network to process textual information such as "red", "green" and "blue". NeuroSolutions for Excel can handle this case by translating this textual information into columns of 0's and 1's.
  • Clean Data
    Neural networks require that you have some numerical value for every input and desired output of every row. Often with real-world data there are missing values, leaving empty or invalid cells in your Excel spreadsheet. NeuroSolutions for Excel will allow you to fill in those cells with an interpolated value, or perform one of several other calculations to clean the data.

Tag the Data

Before training a neural network you need to partition the columns and the rows of your spreadsheet. Each column used by the neural network can either be tagged as an input or a desired output. The rows are then segmented into one or more of the following four groups:
  • Training - Data used by the neural network to learn from.
  • Cross Validation - Data used to evaluate the performance during the learning process to avoid over-training.
  • Testing - Data used to evaluate the performance after the training is complete.
  • Production - Input data to feed into the trained neural network to produce an output.

Create the Neural Network

Once you have selected the data you want to feed to the neural network, you must specify the type of neural network you want to use to model that data. NeuroSolutions for Excel is integrated with the NeuralBuilder, one of the neural network construction utilities included with NeuroSolutions. Once you choose one of the 20+ neural architecture variants available, you can customize the network parameters such as the number of hidden layers, the number of processing elements and the learning algorithm. If you are not sure which parameters to change then don't change any -- the default settings will always give you a working neural network to start with. You can always go back to NeuroSolutions and adjust the parameters later.

Train the Neural Network

Next you tell NeuroSolutions for Excel how may epochs (passes through the data) that you want to train for and let NeuroSolutions do the work of training the network weights. Once the training is complete, a report similar to the one below will be generated within Excel. It gives you the learning curves for both the training and cross validation data sets as well as the mean squared error (MSE) achieved during the training. The network weights are automatically saved at the point where the cross validation error is at is lowest. For the example below, the weights were saved after epoch number 480.

Test the Network Performance

After training you will want to see how well the neural network performs on data that it has not yet "seen". NeuroSolutions for Excel will automatically feed the data that you tagged as "Testing" into the trained neural network and then generate one of the two types of reports shown below. Classification problems are those where the goal is to label each input with a specified classification. A simple example of a classification problem is one where the goal of the neural network is to label each person as male or female (the two classes). In this type of report you are most concerned with the percentage of cases that the neural network classified correctly. Regression problems are those where the goal is to determine a numeric value given a set of inputs. An example of a regression problem is one where you are trying to have the neural network predict a future stock price. Here you are most concerned with how close the network output is to the desired output.

Put the Neural Network to Use

Once you have trained the neural network and tested its performance to make sure it meets your requirements, the final step is to put the neural network to use. You simply enter in one or more rows of new input data and let NeuroSolutions for Excel feed those inputs into the trained neural network and generate an output. The figure below shows an example of a neural network trained to predict the value of next week's S&P 500 index given four attributes of the current week's markets. The four inputs are entered into the last row of the spreadsheet and tagged as "Production", then the system calculates the predicted output and fills in the missing cell.

Try out NeuroSolutions for Excel

Hopefully this brief tour of NeuroSolutions for Excel gave you a little bit of an idea of what the product has to offer. Since this presentation did not cover any of the more advanced features, we recommend that you download a free evaluation copy in order to gain a full understanding of the software. The evaluation edition of NeuroSolutions for Excel does step you through several examples in more detail. In addition, it allows you to design, train and test a neural network with your own data! Click the download button below to obtain your copy today.

Feel free to contact one of our sales associates at 1-800-ND-IDEAS or sales@nd.com if you have any questions on the capabilities of the software or if you would like to place an order over the phone instead of using the secure online order form. Please note that if you should happen to change your mind within 30 days of purchase, you can return the software for a full refund (less shipping charges). Thank you for taking the time to take the NeuroSolutions for Excel tour. Please browse through the rest of the site to learn more about NeuroDimension's products and services.

NeuroSolutions for Excel Product Summary

Click to view screen shot.
NeuroSolutions for Excel is a Microsoft Excel add-in that simplifies and enhances the process of getting data into and out of a NeuroSolutions neural network. This tool benefits both the novice and the advanced neural network developer by offering easy to use, yet extremely powerful features. The foremost feature of this product is that all tasks can be performed directly from Excel!
For the novice, NeuroSolutions for Excel provides:
  • Visual Data Selection
  • One-step Training and Testing
  • Neural Network Design Wizard
  • Automated Report Generation
For the advanced user, NeuroSolutions for Excel has the following features:
  • Data Preprocessing
  • Sensitivity Analysis
  • Data Analysis
  • Batch Experiments
  • Parameter Optimization
NeuroSolutions for Excel is organized into seven modules, each of which can be extended with user-defined custom batches written in Visual Basic for Applications.
Preprocess Data Module
The Preprocess Data module allows the user to easily apply various preprocessing techniques to their raw data to prepare it for input into a neural network. The following Preprocess Data operations are built into NeuroSolutions for Excel™
  • Difference between Columns
  • Sub-sample Rows
  • Insert Column Labels
  • Moving Average of Column
  • Shift - NEW in NeuroSolutions v5.0!
    The input data is adjusted to either move the inputs back by a specified shift value to do predictions or move the inputs forward to lead your desired output.
  • Encode Two Class Column - NEW in NeuroSolutions v5.0!
    The selected column of data is checked to verify that there are two classes contained within the column and is then encoded into another column. The data to be encoded can be textual or numeric, but must be translated to only numeric, integer codes. The encoded column will be written in the first empty column in the dataset.
  • Pause Training - NEW in NeuroSolutions v5.0!
    As an alternative to completely stopping a training run, the network can be paused, so training can be resumed at the same point.
  • Classification Report - NEW in NeuroSolutions v5.0!
    A classification report generated if the desired output of the testing set is a single two-class column.
  • Randomize Rows
    Randomly arranges the rows of data within the active worksheet. This is performed on ordered data so that the training data is representative of the entire data set.
  • Translate Symbolic Columns
    Translates textual columns into columns of 0's and 1's. This is a requirement for columns containing words since neural networks can only work with numeric data.
  • Clean Data
    Cleans the data by replacing blank cells, error codes, and/or user-defined text with an interpolated value, the column average, a random value, or the closest value in a column.
Analyze Data Module
The Analyze Data module provides the user with useful information about their data. The operations available in this module can be used during the preprocessing stage of neural network design or to analyze the network output. The following Analyze Data operations are built into NeuroSolutions for Excel™:
  • Histogram
  • Time Series Plot
  • X-Y Scatter Plot
  • Correlation
    Computes the correlation between each of the columns of data and generates a table.
  • Summary Statistics
    Computes several key statistics for the selected column(s) of data, such as Mean, Standard Error (of the mean), Median, Mode, Standard Deviation, Variance, Kurtosis, Skewness, Range, Minimum, Maximum, Sum, Count, Mean Confidence Level, Kth Largest Value and Kth Smallest Value.
  • Trend Accuracy
    The trend accuracy measure is useful when working with time series data. It gives the percentage for which the actual output changed in the correct direction relative to the previous desired value.
Tag Data Module
The Tag Data module provides a simple graphical method for tagging portions of your data as Training Input, Training Desired, Cross Validation Input, Cross Validation Desired, Testing Input, Testing Desired, and Production Input. This module also provides powerful auto-tag methods. The following Tag Data operations are built into NeuroSolutions for Excel™:
  • Tag Selected Column(s) As Input, Desired or Symbol
  • Tag Selected Row(s) As Training, Cross Validation, Testing or Production
  • Clear Selected Tags
  • Tag Rows By Percentages
    Tags the rows of data within the active worksheet as Training, Cross Validation, and Testing according to user-defined percentages.
  • Select Cross-section
    Allows you to automatically select a cross-section of data based on how it is tagged.
Create/Open Network Module
The Create Data Files module creates tab-delimited ASCII files for each tagged cross-section so that NeuroSolutions can read the data. This module is not used very often since the files are created automatically whenever the neural network is trained or tested. The following Create Data Files operations are built into NeuroSolutions for Excel:
  • Create All Data Files
  • Create Files for one Data Set (Training, Cross Validation, Testing or Production)
  • View Data Files
  • Delete All Data Files
Train Network Module
The Train Network module gives the user the ability to train a network once, multiple times with different random initial conditions, or multiple times while varying one or more network parameters. This powerful module permits the user to easily find the optimum network for a particular problem. The following Train Network operations are built into NeuroSolutions for Excel™:
  • Train
    Trains the active NeuroSolutions breadboard one time and creates a report of the results.
  • Train N Times
    Trains the active NeuroSolutions breadboard N times with different random initial conditions and creates report of the results.
  • Leave N Out Training - NEW in NeuroSolutions v5.0!
    Trains the network multiple times, each time omitting a different subset of the data and using that subset for testing. The outputs from each tested subset are combined into one testing report and the model is trained one additional time using all of the data.
  • Vary a Parameter
    Trains the active NeuroSolutions breadboard N times for each value of a network parameter as the parameter is varied from a user defined starting value by a user defined increment for a user defined number of variations.
  • Train Genetic
    Trains the active NeuroSolutions breadboard while genetically optimizing the choice of inputs and parameter values to achieve the best model.
Test Network Module
The Test Network module can be used to test a network after training has been completed. In testing the network, various performance measures are computed. This module also allows you to perform sensitivity analysis on the network. The following Test Network operations are built into NeuroSolutions for Excel™:
  • Test
    Tests the active NeuroSolutions breadboard on the chosen data set and creates a report of the results.
  • Sensitivity About the Mean
    Performs sensitivity analysis on the chosen data set. This procedure allows you to determine the effect each of the inputs has on the network output
Custom Batches
The user can also create his/her own custom subroutines (batches) for any of the seven modules by calling built-in NeuroSolutions functions and/or writing Visual Basic code. These custom batches can then be run from the NeuroSolutions for Excel™ menu from within Microsoft Excel™ .
For example, suppose that you wanted to create a batch that will train your data using four different type of neural network architectures and then return the results of the architecture that performed the best. NeuroSolutions for Excel will allow you to write a custom VB subroutine to perform these steps automatically and then add that batch as an option within the Train Network Module.
System Requirements:

  • Windows 98/ME/2000/XP/Vista
  • Excel 97 / 2000 / XP / 2003 / 2007
  • NeuroSolutions v5.XX