Skip to main content

Sensor Selection

The Sensor Selection section allows you to evaluate and select the optimal sensors and data channels for your project. It provides tools for filtering, analyzing, and testing various sensor combinations to ensure the best performance and accuracy for your application. This section includes options to view sensor selection results, create new tests, run AI explorations, and analyze confusion matrices to assess classification performance. It helps streamline the process of identifying the most effective sensors based on your data sample list and project requirements.

Data Sample List Tab​

In this tab, you can select the desired Data Sample List for your project. The following information is displayed for each list:

  • Lock Icon: Indicates whether the list is locked or editable.
  • List Name: The name of the data sample list.
  • List Type: Specifies the type of data sample list.
  • Data Shape: Displays the structure of the data.
  • Sample Rate: The rate at which data samples are collected.
  • Number of Samples (n samples): The total number of data samples in the list.
  • Target Range: The range of values for the target variable.
  • Created Date and Time: The date and time when the list was created.
  • Modified Date and Time: The date and time when the list was last modified.
  • Comments: Any notes or comments associated with the list.
  • Status: The current status of the data sample list.

Use the Filter Icon to filter the data sample list by:

  • Name
  • List Type
  • Date Created
  • Data Shape
  • Sample Rate

Click Apply to apply the filter or Clear to reset the filter.

Data Sample List Coverage​

The Coverage section shows the following information for the selected data sample list:

  • Classes: The classes in the dataset.
  • Count: The number of samples in each class.
  • Percentage: The percentage of the total data corresponding to each class.
  • Data Channels: The available data channels for the selected list.
  • Sample Rate: The sample rate associated with the selected data sample list.
NOTE

If your class sample counts are unbalanced, the reliability of the results may be impacted. To address this, the system automatically selects a balanced subset of data for exploration.

Sensor Selection Results​

Under the Sensor Selection Results section, the following details are displayed for the selected data sample list:

  • Favourites: Shows any sensors that have been marked as favourites.
  • Sensor Channels: Lists the sensor channels included in the selection.
  • Overall Accuracy (k-fold): Displays the overall accuracy percentage based on k-fold cross-validation.
  • Accuracy and Precision: Shows the accuracy and precision percentage for each class.
  • Result ID: The identifier associated with the sensor selection result.
  • Create AI Explore: Option to initiate an AI Explore for the selected result.
  • confusion matrix: Displays the confusion matrix for analyzing classification performance.
  • Features Used: Lists the features used in the sensor selection.

Creating a New Test​

To create a new test, click the + icon located at the top right of the Sensor Selection Results section. This opens the Start Sensor Selection window. Here you can:

  • Select Data Channels: Choose which data channels to include in the test.
  • Select Subsets: Specify subsets to test for the best combination of sensors.
  • Sensor Groups Toggle: Use this toggle to view channel groups, or uncheck it to select individual data channels.
  • Test Subsets: Select the subsets of data to test.
  • Test Combinations: The total number of test combinations is displayed at the bottom right of the window.

Click the Start button to begin the sensor selection process. A progress bar will appear showing the progress of the sensor selection. Click Stop Explore to terminate the process.

Managing Sensor Selections​

Once the sensor selection is created, you can perform the following actions:

  • Star: Mark the selection as a favourite.
  • Filter: Filter the results based on specific criteria.
  • Delete: Remove the sensor selection.

These options are available at the top right of the Sensor Selection Results section.

Running AI Explore and Viewing confusion matrix​

  • To run an AI Explore, click the light bulb icon in the Create AI Explore column. This initiates an AI-driven exploration of the selected result.
  • To view the confusion matrix, click the corresponding link to access the K-fold and training separation confusion matrix. This matrix shows class separation performance along with the overall accuracy and F1 score.