Specs

In Uncountable, a spec is a set of goals you define for test outputs in a project (for example, wanting viscosity between 300 and 500). It reflects the outputs you want to optimize when modeling or creating new experiments.

Specs can be set at either the project level or individual recipe level, and they’re used for several purposes:

• Optimizing and modeling outputs you care about
• Setting upper/lower limits and goals for project success
• Highlighting values in measurement pages (red for values not meeting limits, green for values meeting limits).

Specs are essential for features like Objective Search and visualization indicators, and it’s recommended to create them early in a project’s lifecycle.

Creating a Project Spec

To set up a new spec from scratch, select “Calculate” from the Uncountable navigation bar and click “Set Spec”.

Specs can also be accessed from a project’s information header.

When accessing a spec for your project, you will either see the project’s default spec or a blank page if no spec has been created or set as the default.

At the top of the page, you can configure the spec name, material family, and project (A). Below, a table allows you to define spec goals—the target outcomes for your project (B). The spec table has two tabs:

• Goals: Manually configure each component of a spec goal, including the goal (output or calculation), priority, spec type, and settings.
• Goals Listing: Select from a list of previously configured goals, making it easy to reuse goals across multiple specs

Adding Goals

Within the Goals tab, add the desired outputs or calculations by selecting from the options below:

• Add Outputs: Manually select outputs to include in your spec. You can also use the button menu to select from the following options:
  • Add Project Outputs: Adds outputs without any conditions attached.
  • Add Project Outputs Including Conditions: This option will add outputs with all present conditions.
  • Add Common Outputs: Adds the 1-20 most commonly used outputs in the project, excluding condition parameters.
• Add Groups: Add output groups to better organize your outputs. Learn more about Output Groups.
• Add Calculations: For output calculations, ensure that they are active and part of the project in order to be included in the spec. You can also add input calculations (such as formulation cost) for additional context and accuracy. Learn more about Output Calculations.

Configuring Goal Conditions

Clicking on an output in the spec opens a dropdown menu. By selecting Configure Condition(s), users can control how condition parameters—such as test methods, test method, or aging time—are used in modeling. You can either focus on specific conditions or combine all available data, depending on how much these factors influence your results.

Consider this example scenario:

On Enter Measurements, a recipe has three recorded density values under different test methods:

• Method A: 0.6
• Method B: 0.7
• Method C: 0.95

On the Specs page, the spec for density has been configured to include only Method A and Method B, meaning:

• The modeling process will use only the 0.6 (Method A) and 0.7 (Method B) values.
• The 0.95 (Method C) measurement will be ignored because it wasn’t selected in the Configure Condition(s) dropdown.

This approach is useful when you know certain condition parameters significantly impact results and want to refine the dataset accordingly. However, filtering conditions may result in a smaller dataset.

Alternatively, if no conditions are selected:

All condition parameters will be collapsed into one, meaning the model will use the average of all three values:

This method works well when condition parameters have minimal impact, allowing for a larger, more statistically robust dataset.

Additional Goal Settings

The goal settings menu also includes the following options:

• View Output Details: Redirects you to the “Edit Measurements” page for the selected output.
• Swap Output: Allows you to replace the output with another while preserving its priority, spec type, and goals.
• Configure Relative Property: Allows the user to set a relative spec, a dynamic specification that automatically updates based on theoretical calculations. Instead of using fixed values, relative specs adjust their ranges automatically when formula changes occur.
• Use Log Scale: Switching to a log scale helps when modeling properties that vary widely, like viscosity. It adjusts the goal to a logarithmic format, making it easier to handle large value differences and improve optimization.
• Copy Goal: Duplicates the goal to allow users to change output conditions for a different goal search.
• Delete Goal: Removes the current goal from the spec.

Setting Priority

Priority determines an output’s importance in the optimization process. You can choose from:

• Ignore: Excluded from optimization (equivalent to not adding it to the spec).
• Low: Included but not a primary focus.
• Medium: Given considerable attention.
• High: Critical to success (limit to 1-3 per spec).

Note: When using Analyze Experiments with AI, you can include as many outputs as needed. However, for Suggest Experiments with AI, it’s best to optimize fewer than ten at a time. If optimizing many outputs, prioritize a subset first, then refine additional outputs in later rounds.

Clicking on the Priority header allows users to apply a selected setting to all goals at once.

Setting Spec Type

Spec Type defines how an output should be optimized:

• Maximize: Aim for the highest possible value.
• Minimize: Aim for the lowest possible value.
• Range: Keep values within a set range.
• Target: Aim for a specific value.
• Range with Target: Define both a range and an ideal target value.
• Range with Target Range: Define both a range and an ideal target range.
• Exists: Ensures the output is present when measured.

For categorical outputs (i.e. pass/fail), the available spec types are:

• Allowed Options: Specify preferred values.
• Disallowed Options: Exclude unwanted values.
• Multiclass: Used only for Analyze Experiments with AI to apply a classification model.
• Exists: Ensures the output is present.

Clicking on the Spec Type header allows users to apply a selected setting to all goals at once.

Setting Goal Limits

Under the Settings column, enter the upper and/or lower limits for each output to define success criteria. These limits, often based on specification sheets, determine how values are highlighted on the Enter Measurements and Compare pages:

• Red: Outside the limit.
• Green: Within the limit.

Considerations When Setting Limits

1. What are the true requirements of this project?

Your limits should align with the project’s core objectives, which may stem from:

• Customer requests (e.g., improving tensile strength or elongation).
• Internal goals (e.g., reducing formulation costs by 10% while maintaining performance).
• Research objectives (e.g., identifying the best-performing polymer before and after aging tests).

2. What does historical data reveal?

Understanding past data helps set realistic and informed limits. Click on the histogram bar next to a goal to access a dropdown with statistical insights, including:

• Mean value & standard deviation
• Histogram distribution
• Goal-to-threshold distance (if limits are set)

This data provides a quick reference to past experiments, indicating what has been achievable and what may be difficult. You can also enable “Show Data Output Project” to view data across your entire material family.

3. Are there trade-offs between outputs?

Many outputs are correlated, meaning optimizing one may impact another. Click the View Tradeoffs button (bottom-right) to navigate to the View Correlations page.

For example, if tensile strength and viscosity have a strong positive correlation (i.e. 0.93), but your goal is to maximize tensile strength while minimizing viscosity, achieving both may be challenging. Breaking these trade-offs typically requires:

• New ingredients or process parameters
• Exploring entirely new input spaces

Saving, Loading, and Locking a Spec

Once you’ve filled a spec with the necessary outputs and calculations, you can save it for future use in Analyze Experiments with AI or Suggest Experiments with AI.

Saving a Spec

To save your spec:

• Click “Overwrite Spec” to update the existing spec.
• Click “Save As…” to create a new version with a different name.
• Set the spec as the project default if needed.

You can view all available specs under Design > Specifications.

Loading a Spec

If you need to modify an existing spec, loading and adjusting it can save time compared to creating one from scratch. To do this:

1. Open the desired spec.
2. Make the necessary changes.
3. Click “Save As” to save it under a new name if needed.

You can also load specs from past projects or other users’ projects:

• Click “Load Spec” on the right side of the page to open the selection window.
• Choose from:
  • Load Spec: Select from specs saved under the current project.
  • Import Spec: Access specs saved under the current material family.
  • Load Recipe as Base: Import all outputs from a specific recipe into the spec.

Locking a Spec

To prevent further modifications, lock a spec by clicking the lock button next to the spec name. Once locked all fields under Priority, Spec Type, and Settings will be grayed out. No one will be able to edit the spec until the lock has been removed.

Displaying Specs on Enter Measurements

You can choose to display specs permanently while entering data on the Enter Measurement page.

Displaying a Single Spec

If you have created a spec for your project, it will automatically display on the Enter Measurements and Compare pages, highlighting outputs with defined specs. To add a Specs row to your datagrid, use the Search Action bar to search for and add “Specs Row”.

Edit specs directly on the Enter Measurements page by clicking into the cells. This opens the Measurement Details modal, within which you can manually adjust spec settings.

Displaying Multiple Specs

If your experiments have different assigned specs, use the Assigned Specs option:

• In the Specs row, click the column header for an experiment.
• Select Assign Specs to choose a different spec for the experiment.

This is especially useful when comparing recipes in a production environment. For more details, see Assigned Specs on Individual Recipes.