Understanding usage of Volvo trucks - Revision history

Slawek at 11:53, 16 October 2018

2018-10-16T11:53:17Z

Slawek at 11:52, 16 October 2018

2018-10-16T11:52:07Z

Slawek at 18:11, 14 October 2018

2018-10-14T18:11:33Z

Slawek: Created page with "{{StudentProjectTemplate |Summary=A project in collaboration with Volvo AB on understanding vehicle usage patterns. |Keywords=Data Mining. Data representation |Prerequisites=G..."

2018-10-14T18:11:20Z

Created page with "{{StudentProjectTemplate |Summary=A project in collaboration with Volvo AB on understanding vehicle usage patterns. |Keywords=Data Mining. Data representation |Prerequisites=G..."

New page

{{StudentProjectTemplate
|Summary=A project in collaboration with Volvo AB on understanding vehicle usage patterns.
|Keywords=Data Mining. Data representation
|Prerequisites=Good knowledge of machine learning and programming skills for implementing machine learning algorithms
|Supervisor=Sławomir Nowaczyk, Fredrik Moeschlin (at Volvo)
|Level=Master
|Status=Open
}}
In discussions different stakeholders pointed out that the current segmentation of Volvo customers is not optimized. There is a large value in correctly understanding and grouping customers' needs in order to optimize offerings and sales. We are currently looking into the segmentation as it is today and it is clear that we can utilize techniques from data mining and machine learning applied on telematics data to better cluster vehicles according to their usage. On a high level we would expect segments such as “City distribution”, “Regional distribution”, … . They are currently described out of their usage, for instance Interregional Haul:

“Long distance transport followed by few clustered deliveries in normally smooth but occasionally also in rougher road conditions. The average distance between delivery and pickup is between 50 km and 250 km. The vehicle often returns to its home base during the day, which means that overnight stays in the vehicle occur in average 1-2 times per week.”

From telematics systems we have detailed information about where users drive, times, distances, locations, different events (stop, login, Power Take-Off (PTO), etc.). If we combine this information from several vehicles and from map data sources (e.g. Open Street Map), we could generate detailed insights of how the vehicle is used. This can lead to more precise segmentation (clustering) with better quantified attributes (road conditions, driving time, actual hub-hub distances etc). We would also be able to generate a more granular segmentation, or sub-groups within the segments. Furthermore we can give an overview of the segments in terms of quantity of customers, regional variation and more. The core value for Volvo is better market understanding to allow for better positioning and offerings.

@@ Line 8: / Line 8: @@
 |Status=Open
 }}
 There is challenge in the automotive industries to properly segment offerings and products. By correctly understanding and grouping customers' needs, offerings and sales can be optimized. By utilizing techniques from data mining and machine learning, applied on telematics data,  vehicles can be clustered according to their usage. On a high level segments can be described as e.g. “City distribution”, “Regional distribution”. These groups can then be further described for instance out of the distances, usage patterns, operating domain and other factors.

@@ Line 8: / Line 8: @@
 |Status=Open
 }}
-In discussions different stakeholders pointed out that the current segmentation of Volvo customers is not optimized. There is a large value in correctly understanding and grouping customers' needs in order to optimize offerings and sales. We are currently looking into the segmentation as it is today and it is clear that we can utilize techniques from data mining and machine learning applied on telematics data to better cluster vehicles according to their usage. On a high level we would expect segments such as “City distribution”, “Regional distribution”, … . They are currently described out of their usage, for instance Interregional Haul:
+There is challenge in the automotive industries to properly segment offerings and products. By correctly understanding and grouping customers' needs, offerings and sales can be optimized. By utilizing techniques from data mining and machine learning, applied on telematics data,  vehicles can be clustered according to their usage. On a high level segments can be described as e.g. “City distribution”, “Regional distribution”. These groups can then be further described for instance out of the distances, usage patterns, operating domain and other factors.
-“Long distance transport followed by few clustered deliveries in normally smooth but occasionally also in rougher road conditions. The average distance between delivery and pickup is between 50 km and 250 km. The vehicle often returns to its home base during the day, which means that overnight stays in the vehicle occur in average 1-2 times per week.”
+From telematics systems, detailed information can be attained on where users drive, times, distances, locations and different activities. If we combine this information from several vehicles and from map data sources (e.g. Open Street Map), we could generate detailed insights of how a vehicle is used. This may lead to more precise segmentation (clustering) with quantified attributes (road conditions, driving time, actual hub to hub distances etc.). We may also be able to generate a more granular segmentation, or sub-groups within the segments. Furthermore we expect to enable an overview of the segments in terms of quantity of customers, regional variation and more. The core value to reach for is better market understanding to allow for better positioning and offerings.
-From telematics systems we have detailed information about where users drive, times, distances, locations, different events (stop, login, Power Take-Off (PTO), etc.). If we combine this information from several vehicles and from map data sources (e.g. Open Street Map), we could generate detailed insights of how the vehicle is used. This can lead to more precise segmentation (clustering) with better quantified attributes (road conditions, driving time, actual hub-hub distances etc). We would also be able to generate a more granular segmentation, or sub-groups within the segments. Furthermore we can give an overview of the segments in terms of quantity of customers, regional variation and more. The core value for Volvo is better market understanding to allow for better positioning and offerings.
+From telematics systems, detailed information can be attained on where users drive, times, distances, locations and different activities. If we combine this information from several vehicles and from map data sources (e.g. Open Street Map), we could generate detailed insights of how a vehicle is used. This may lead to more precise segmentation (clustering) with quantified attributes (road conditions, driving time, actual hub to hub distances etc.). We may also be able to generate a more granular segmentation, or sub-groups within the segments. Furthermore we expect to enable an overview of the segments in terms of quantity of customers, regional variation and more. The core value to reach for is better market understanding to allow for better positioning and offerings.

@@ Line 2: / Line 2: @@
 |Summary=A project in collaboration with Volvo AB on understanding vehicle usage patterns.
 |Keywords=Data Mining. Data representation
 |Prerequisites=Good knowledge of machine learning and programming skills for implementing machine learning algorithms
 |Supervisor=Sławomir Nowaczyk, Fredrik Moeschlin (at Volvo)