How Machine Learning Improved a Car Production with AI Predictions
Clearly, increasing efficiency with the lowest possible consumption of materials is the biggest challenge for car factory owners. Read how we use machine learning to help a leading brand on the global automotive market to improve the prediction of materials consumption.
Our client, a German car manufacturer, want to predict daily glue consumption based on the type and number of parts that enter the factory on a day. The purpose of daily consumption prediction is to improve the planning of glue stores and save unnecessary purchases.
First, the client turn to the data scientists team from Stermedia to verify the idea of the internal team and refine the business side of the project. And, without a doubt, conduct analysis using artificial intelligence (machine learning).
At first, our data scientists conduct a full business interview to match the most adequate solution. Secondly, they collect information about how the factory works, what data the customer has, and when data is collected. Next, we create a model for data analysis.
Definitely, the challenge that is encountered is the client has only a small amount of data. For this reason, regularize linear models with feature elimination are used in analytical work. For each model that we are building, we go through iterations and repeat the following steps (we start with all the car parts amounts as features):
- Cross validate (check the performance of) the model with the car parts data that we still didn’t remove. Secondly, remove the part that got the lowest ratio (“relevance score” from the model)
- Looking at the average cross-validation errors perform on smaller and smaller sets of parameters. We decide on the optimal number of parts that enter the model. For each value describe – with the help of cross-validation and correlation analysis – features were developed. Finally, that allows the value to be described best and build a suitable model.
Pic. 1 A set of test data to check the model
In the chart on the left (chart of expecting consumption relative to the real one) the closer the dots are, the better. On the right, blue is real adhesive consumption, and yellow is expecting consumption. They are sort so that you can see how high the values are. This means that a significant amount of glue is used.
Furthermore, for more data, the team would consider using random forest or simple neural networks. With huge numbers of data points, even more complex neural architectures or gradient boosting methods like XGBoost or Light GBM would be use.
As a result, we improved the quality of materials consumption prediction by 15%. According to the weight mean absolute percentage error. In a longer perspective, this prediction leads to more accurate planning and waste reduction in the manufacturing process.
In the above work, the model use jupyter environment for Python and the scikit learn and pandas libraries.