Predii Safety Analytics™
Which parts and components will fail? When (at what mileage) will they fail? What impact will these failures have? And: what needs to be done to avoid it?
Predii Safety Analytics™ can help answer these questions for Automotive Manufacturers, Product Managers, and Quality Engineers. Our Natural Language Processing engine is able to extract predictive and prescriptive insights from customer verbatim data (e.g. found in customer complaints), symptoms and failures originating from Repairs Orders, or connected car/telematics data.
Predii Safety Analytics™
Example Application NHTSA Safety Data
The Predii Insights Dashboard - specifically fine-tuned to answer safety and quality related questions - offers an aggregated view of extracted insights on a Year, Make, Model, Engine and even Parts level.
These insights can include:
Top Customer reported concerns ("symptoms")
Top Technician reported failures
Top replaced parts & parts failure rates
The National Highway Traffic Safety Administration (NHTSA) maintains public data base for consumer reported safety issues. NHTSA uses this information as a basis for investigation into defects and potential vehicle safety recalls. Predii Safety Analytics Beta extracts aggregated insights from this data as shown in the example below.
Interested in a full demo? Contact us!
Vehicle Make and Year Filter
In total, the NHTSA data base contains more than 1.2 Million records on real life vehicle issues, safety & quality concerns, and component failure. This POC Dashboard is based on 9,656 (customer-reported) extracted symptoms and 11,299 (technician-reported) failures across 268 makes¹ and 6,138 models.
All Predii Insight Dashboard allows viewers to filter for Makes, Models, and Vehicle Year for refinement. Our enterprise solution dashboards additionally included Geo Location and Data Origin filters.
¹total of extracted makes: 1056 incl. trucks, trailers, RV. For quality of the analysis we considered all makes with records >10
Top (Driver-reported) Customer Concerns
'Customer verbatim’ data is extremely hard to translate into usable analytics. Customer complaints consist of textual data, real life human language, which is unstructured, messy, and contains very domain-specific automotive terminology.
Predii’s Natural Language Processing engine has been purpose-built to extract predictive and prescriptive insights from unstructured textual, sensor, and procedural automotive data; our algorithm has been fine-tuned to understand automotive jargon for 8+ years.
This allows us to translate the entirety of individual records into classified, categorized symptoms as shown in the example.
Top (Technician-reported) Vehicles Failures
In additional to verbatim data ('technician notes)', Technician-reported failures contain Diagnostic Trouble Cdes, Part numbers, ACES/PIE definitions, and more. The Predii technology is able to capture thee data points, categorize them, and translate them into actionable insights.
Top Technician Reported Failures for this POC data set include: Check Engine Light, specifically defined noises heard from different parts of the vehicle ("knocking", "grinding", "clicking").
Top Contributing Makes & Models
The Predii Insights Dashboard does not only show aggregated views across all symptoms and failures but allows viewers to narrow down to specific Makes and Models contributing to these symptoms and failures.
The Enterprise version of our Dashboard additionally includes Vehicle Year and Engine filters.
Odometer readings show the mileage distribution for each concern/failure per make and model, allowing users to identify patterns and trends. Mileage distribution is crucial for advanced analytics including prediction of quality issues.
Curious to learn more about the history of our NHTSA data POC? Check out our blog!