The customer identified logistics as a major cost factor in its production chain that could be optimized. A solution was sought that would minimize the necessary transportation activities in the target market. There, vehicles are transported to the dealers centrally, but not in a data-driven manner.
For the POC we decided for a setup where provided data was handled from within a secured local environment. After deep analysis of market properties and correlations, Contiamo compared multiple algorithms. The final artificial intelligence contained a classification model that optimizes the distribution of cars within the target market. It receives multiple inputs like sales data, current stock data, market demand and dealer information. The output format was selected to be optimally suited for evaluation and performing further analyses about transportation activities. The setup of the pipeline was created to be easily transferable to the client’s cloud environment which was prepared in parallel.
For this project, Contiamo and the client decided to work in weekly sprints. After initial data exploration and checking suitability of the data, statistical analyses on transportation activities were performed. Results from the investigations were iteratively discussed with the customer. Three promising AI algorithms were identified to answer the initial question: addressing different targets following either regression or classification algorithms. The preliminary models were compared in terms of their correctness and informative value. The customer decided on a model according to our recommendations. This was subsequently improved and refined. This PoC model is able to accurately predict the success of logistics activities based on historical data. Results are to be discussed by the customer now.
Reduce logistics activities by 40%
75% accurate prediction of sales locations
Distribution optimization
We created a machine learning model to reduce logistics movements of cars by predicting likely sales patterns to identify which cars to move and when. Using efficient and state-of-the-art libraries, a powerful artificial intelligence was created with a performance of 75% matching historical sales recommendations. The high accuracy provides an excellent basis for a comprehensive optimization of logistics taking into account various other market conditions (e.g. consistent availability, dealer demands, ordering processes) in the next step.
.png){kind=small}
.png)
.png)
.png)