> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.umnai.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.umnai.com/_mcp/server.

# Strengths and weaknesses

Strengths and weaknesses, or SAW, describe confidence-related behaviour in an Umnai model.

SAW provides two measures: strength and uncertainty. Together, they help explain where the model is supported by strong evidence and where its behaviour may be less certain.

## Strength

Strength reflects how strong the model is in the activated [partitions](../../explainable-neural-net-xnn#partition).

It is based on:

| Factor                | Description                                                                 |
| --------------------- | --------------------------------------------------------------------------- |
| Training data support | The amount of training data in each activated partition.                    |
| Partition model fit   | The quality of the partition's polynomial model, or `then_expression`, fit. |

A stronger partition is one where the model has more support from training data and a better local fit.

## Uncertainty

Uncertainty reflects how certain the model is about its prediction behaviour.

It is based on:

| Factor                        | Description                                                                |
| ----------------------------- | -------------------------------------------------------------------------- |
| Estimator variance            | Variation across estimators.                                               |
| Partition prediction variance | Variation in the predictions produced by the partition's polynomial model. |

Higher uncertainty indicates that the model behaviour is less stable or less certain for the activated partitions.

## Aggregation

Strength and uncertainty can be aggregated across activated partitions.

This allows SAW to produce confidence estimates at different levels:

| Level                | Description                                                |
| -------------------- | ---------------------------------------------------------- |
| Data segment         | Confidence estimate for a selected segment of data.        |
| Batch of predictions | Confidence estimate across a batch of prediction requests. |
| Dataset              | Confidence estimate across the dataset as a whole.         |

At a global level, these aggregations help identify where the model is generally strong, where it is uncertain, and which areas may deserve further review.