Making a neural network with data given.

So I have voltage, current and power data. They all are related with the equation P = VI. The data represents the operation of a fuel cell. It has three different operations. 1. Normal Operation 2. Purging and 3. No load operation. I have to build a neural network that can predict or tell what operations is happening when it is fed with data.

Im a newbie to machine learning. Any help will be really appreciated.

Answer 1

[Ritvik Garg]

Hi Bhavick,

This is a Machine Learning Classification Problem with input variables as voltage, current and power data, and output variable as operations. You can write your code in MATLAB or use toolboxes like Deep Learning Toolbox which provide command-line functions and apps for creating, training, and simulating shallow neural networks. The apps make it easy to develop neural networks for tasks such as classification, regression (including time-series regression), and clustering. After creating your networks in these tools, you can automatically generate MATLAB code to capture your work and automate tasks.

As with function fitting, there are two ways to solve this problem:

• Use the nprtool GUI, as described in Using the Neural Network Pattern Recognition App.
• Use a command-line solution, as described in Using Command-Line Functions.

You may refer to example-1 and example-2 for classification problems.

[Bhavick Singh]

@Ritvik Garg I have voltage data. In that data there are three states. Normal, No load and Purge. I have to build a machine learning algorithm which can predict what state the voltage is showing at each point when it will be fed with data. Which one will be the best classification algorithm to use for this problem?

Note: I have one input variable which is voltage.

[Ritvik Garg]

Hi Bhavick,

You cannot figure out the best model for your use-case without experimenting different models. If you have large dataset, you may go for trying deep learning models also, or else you may go for traditional machine learning algorithms.

Popular algorithms that can be used for multi-class classification include:

• k-Nearest Neighbors.
• Decision Trees.
• Naive Bayes.
• Random Forest.
• Gradient Boosting.

Try these and let me know which performed best.