Data driven decisions to calculate CO2 emission for materials derived from estimated vehicle weight

Change in the gross weight of the car due to change in Special Accessories (SA)/parts/different materials can drastically alter the overall CO2 emission readings. Each carline has multiple variants offered to the customers. Each variant additionally has different code rules associated with them further increasing the customizability. This requires a system to predict the estimated gross weight of the cars to calculate the CO2 emission accurately. Carlines, which are already in production phase, have certified weights associated with them. For each variant of the car, there is a specific weight threshold, which governs the emission certification of it. However, carlines, which are in development phase, have yet to certify for CO2 emission. This makes it very difficult to identify customizability options for development variants. Any change in the part/accessory can make the variant exceed the weight threshold, thus, making it difficult to offer alternatives. The idea is to help the sustainability team to take data driven decisions rather than relying on their hunch to calculate the CO2 emission for the material/part, which adds up to the delta weight. The basic approach to develop any ML model is to gather proper data. We gather data for the following:  Valid part combination and code rule  Part information with material type and weight Certified weight of the variant is calculated for historical/current variants. This data (including material type) is fed as training set to the ML model. The team responsible for energy efficiency and sustainability use material data along with the predicted delta weight to calculate the CO2 emission precisely. We identify that the ML approach to train the model with carefully collated material, part and weight data can help the sustainability team to accurately predict the violations in estimated weight of the car variants or the threshold breach in CO2 emission. The numerical data generated by the model is employed to develop a BI dashboard to take effective decisions during the development phase of a carline and its variant offerings. Conclusion  The sustainability team can identify the materials contributing more to increase in weight thereby increasing CO2 emission, and further take data driven decisions to reduce it  The model output is used as baseline creation for CO2 target at material level, thus, collating the material versus CO2 data for present and future years  The accurate prediction of the model helps to reduce the turnaround time taken to recertify the car variants in case of breach in the weight threshold