FEM: Outlier Detection Guide for Facilities

Table of Contents

• Accessing Outlier Detection
• What kinds of automated data quality checks are available?
• What can I expect with Outlier Detection?
• What is considered an Outlier?
• FDM to FEM Outlier Detection
• FAQ

High-quality data is the most valuable component of the Higg Index and the Worldly platform. In order to maintain high-quality data, it is important to identify and correct any outliers. 

An Outlier is a data point or calculation that differs significantly from our historical observations and expectations.

Before posting your Assessment, you may see a message that prompts you to review outliers. Click Review Outliers to see which items you need to check.

Accessing Outlier Detection

The Outlier Detection panel will come from the right-side of the screen, which will guide you through reviewing any outliers detected in your assessment.

Under the text, you will see hyperlink(s) prompting you to review your inputs. For example, in the screenshot below you are prompted to Review Finished Product Assembler and Review Production Volume. Clicking on the link will bring you to the part of the assessment where the suspected outlier is.

Outlier detection identifies any outlier, or data point/calculation that is significantly different, that you can dismiss. These are highlighted in yellow.

For example, in this screenshot below, the normalized energy usage appears high for this facility type. Before proceeding, you must review all energy sources for errors including the reported production volume. If you’ve reviewed the data and can confirm it is accurate, check the box next to “I reviewed and it looks correct”. You can also select Review Later to dismiss the outlier.

1. Selecting I reviewed and it looks correct will open a text box where you can add a comment about dismissing the flag.
   
   Clicking Save will move the outlier to the Reviewed list. It will display who marked the outlier correct, as well as the date it was marked correct. Select the arrows next to the outlier to jump to the specific question within the assessment.
   
2. Selecting Review Later will keep the flag on the To Review list. Click the yellow tab to access the list of outliers. 
   

While you’re finishing your assessment, you can click the yellow tab to check anything that still needs to be reviewed. 

The yellow tab turns green once there are no more outliers to be reviewed.

Validations, highlighted in red, identify any errors that you must address and cannot dismiss.

For example, in this screenshot below, the energy mix of the purchased steam for the reporting year needs to add up to 100%. You will not be able to post the assessment until the value is reviewed and changed.

Once you see this message, you can post your assessment. 
NOTE: FDM and FEM require outlier checks before posting.

Click here for more examples of outlier detection and validation in the FAQ. 

What kinds of automated data quality checks are available?

The Worldly platform automatically runs the following data checks during the facility self-assessment process.

Higg FEM Outlier Detection Methodology

To learn more about outlier detection, read Worldly's Higg FEM Outlier Detection Methodology.

Summary: A high-level overview of when outlier detection is triggered and what it means.

Definition of Data Outliers: What is the difference between an erroneous outlier vs a true outlier?

Initial Data Set: Cascale and Worldly used the complete set of FEM23 assessments to establish outlier thresholds for use with FEM24 assessments.

Approach to Identifying Outliers: Learn how a standard statistical method, the Interquartile Range, is used to identify outliers.

Single Year-on-Year (YoY) Change: Read about how the YoY comparison flags anomalously large values and what went into the development of this feature.

Total Energy/Water Outlier Thresholds: View these two tables to understand the threshold values for Energy and Water outliers.

What can I expect with Outlier Detection?

Outlier detection was released on Feb. 19, 2025. If a facility’s Higg FEM assessment was posted before that date, the facility will need to wait until after this date to be able to take advantage of data checking. 

In addition, as long as they have not begun verifying data with a verification body and it is before April 30, 2025, facilities can un-post their assessment to use data check to look for outliers in their responses before reposting their assessment.

For example, if a facility posted their assessment on January 31, 2025, they will need to do the following between February 19 and April 30, 2025: 

1. Un-post their assessment
2. Use outlier detection to check their assessment data
3. Repost their assessment

Year-over-year anomaly detection will also be available in mid-March.

What is considered an Outlier?

• Erroneous outliers are data points that arise from errors during data collection, recording, calculation, or entry. They represent inaccuracies and do not reflect true underlying patterns in the data. For example, a typographical error that records a person’s age as 250 years instead of 25 would be an erroneous outlier, as it does not correspond to a realistic value. Similarly, recording an energy value in the wrong units without conversion (e.g., a value in megajoules, MJ, assigned units of kilowatt-hours, kWh) also leads to erroneous outliers. Such “bad data” must be identified and addressed – typically by removal or correction (e.g., via verification or replacement with imputed values) – to maintain the integrity of the analysis.

• True outliers are legitimate data points that are statistically rare but not erroneous. They occur naturally in the data and reflect real phenomena or occurrences. True outliers often provide valuable information about unusual but valid behaviors or events. For example, an anomalous increase in facility energy use might correspond to increased cooling implemented during a heat wave. True outliers can reveal critical trends, risks, or opportunities and should not be dismissed without analysis.

FDM to FEM Outlier Detection

Both FEM and FEM require outlier detection before posting.

Please note that although FDM data imported to FEM is not checked during the transfer, FEM runs outlier detection on the imported data before the assessment is posted.

FAQ