Customer
-
A global professional consulting business
-
Offering services across multiple sectors such as Marine, Energy and Manufacturing
-
Looking at ways to make the workplace a safer place for their employees
Operations
-
7,000
employees
-
195
offices
-
78
countries
Results
Deep analysis of HSE incidents to derive actionable insights
-
50%
reduction of incidents marked as "Other".
-
7,000
incidents accurately reassigned.
-
60X
resource efficiency. Reducing 4-6 weeks of work into a 4-hour task.
Customer challenge
The HSE team were receiving around 1,000 HSE reports a month, and over the last few years had accumulated circa 78,000 reports. These reports were a mixture of Lost Time Incidents (LTIs), Near-Misses and Observations.
Due to resource constraints, the HSE team were only able to review LTIs, meaning that key data from near-misses and observations (that could potentially prevent a future accident occurring) were not being analysed and actioned.
There was a need to quickly and accurately review the textual data in the HSE reports, and extract information that would assist in shaping HSE strategy and interventions and improving safety.
The human intelligence
LR’s data science team used NLP and machine learning to solve for the challenge posed. First the original datasets were cleaned and accurately reclassified. With this clean and detailed data set, LR’s HSE subject matter experts mapped and refined the risk categories emerging from the data. Data science then helped identify key locations and business units under risk.
The smart solutions
- Clean the original incident data submitted by employees and ensure the user assigned labels were accurate
- Accurate recategorization of incidents tagged as “OTHER” with the help of data analytics and human intelligence
- Deep diagnostics of free text incident descriptions to get deeper insights into the incidents
- Using deep insights to develop nuanced and detailed category classification and enable faster Root Cause Analysis (RCA)
The outcome
This project brought the customer the following benefits:
- The “Other “category initially contained 7,000 incidents, but by using NLP to more accurately reassign incidents to different hazard categories, this was reduced by 50% to 3,500.
- Of the total 78,000 HSE incidents reported over a 5-year period, LR supplemented all 100% of incidents with additional hazard categories, thus speeding up RCA.
- Previously, cleaning a dataset of 78,000 records would have taken the client approximately 4-6 weeks, but by using the NLP solution from LR, this was reduced to 4 hours.
LR was able to provide secondary insights to all 78,000 incidents, thus providing previous unknown details to assist with HSE management.
Bringing data science to safety.
With LR SafetyScanner gain a clearer picture of where your biggest risk lies, and where to accurately target your safety efforts.
Find out more about LR SafetyScannerCASE STUDIES
Who we work with
We help businesses across dozens of sectors push forward and achieve like never before. How can we help you?
INSIGHTS
What we think
LR's experts regularly share their research and insights.
