Key results
Background
For the past four years in France, a major car manufacturer has been using Expleo’s data science expertise, specifically in textual analysis, to obtain an exhaustive overview of customer satisfaction with its vehicles in service. What is the objective? To make the voice of the customer better heard in millions of unstructured data.
The Challenge – Listening to the Customers Voice
Today in the automotive sector, checking a car’s performance is as much a traditional control of mechanical gears as an update of embedded systems. Whether it is to guarantee passenger safety, ensure driver comfort or efficiently inform customers about their journey, the slightest bug is excluded in this multitude of data. To better anticipate what could go wrong with its car ranges, the prominent French car manufacturer has been using Expleo’s expertise in data science, textual analysis, and RAMS (Reliability, Availability, Maintainability & Safety) engineering to conduct an extensive 360° study of all reported quality returns. Faced with thousands of after-sales exchanges every month, analysing this great quantity of data is a challenge. The data includes many satisfaction survey results, workshop or garage feedback, call centre telephone transcripts and opinions shared on social networks or specialist forums.
Solutions – Combining Engineering & Big data analytics
Processing such a large amount of information “manually” is too time-consuming without the help of Big Data tools offering an optimal combination of RAMS analysis, artificial intelligence, and data visualisation capabilities. This is the know-how, that the experts deployed by Expleo on this project provide to our client’s quality department. The team included the RAMS engineers in charge of studying the operating safety of the targeted vehicle ranges; the data scientists specialising in natural language processing to better analyse the numerous written information sources; and the data analysts in charge of working on the design of the data visualisation dashboards. AFor the time being, the solution deployed by Expleo covers our client’s vehicle fleet in France. For this project Expleo relies on the combination of the Python programming language, a data analysis method based on natural language processing (NLP), as well as the TIBCO Spotfire, R Shiny or Dash data visualisation platforms.
Outcome - Improved understanding of customer's feedback
Ehe first significant benefit for our client is a drastic reduction in the time and technical or human resources required to analyse its field reports. For example, several days of manual transcription are reduced to just a few minutes of data processing. Thanks to Expleo’s NLP algorithms sophistication, false alarms about incapacitating breakdowns have been significantly reduced in keyword searches. Following a specific test phase, Expleo’s algorithms can now reduce the readback rate by 50%, with a margin of error of only 5%. The advantage was quickly noticeable for our client’s quality department on a word like ‘traffic light’, whose many possible interpretations (low beam, positional light…) could lead to a lesser relevance in the diagnosis. Beyond this example, the correct diagnosis is, of course, the overall objective. And the results are more than notable there, too, with the combination of data intelligence and data visualisation tools. Thanks to the dashboards set up by Expleo, processes have been automated, and correlations between the information collected can be highlighted faster. Currently, Expleo’s algorithms classify potential quality problems in 600 failure scenarios with an accuracy of at least 80%.
With the automation, comfort, and speed of use of the Expleo solution, our client's quality teams benefit from improved reproducibility of the identification criteria for their field reports. This is invaluable to meeting customer expectations as closely and quickly as possible.
Lead Data Scientist, Expleo
DIGITAL TRANSFORMATION
Big data, analytics, AI & advanced algorithms
Expleo builds modern data analytics strategies by deploying machine learning and advanced algorithms. This empowers us to create complex predictive and governing models, both in the industry and services environments.
