Main failure factors of rocket launches and their impacts

Details

Supervisors

Creemers, Stefan

Faculty

Louvain School of Management

Degree label

Master [120] : ingénieur de gestion, à finalité spécialisée

Abstract

enThis master dissertation aimed to answer the following questions: What are the main failure factors of rocket launches? What are their impacts on the probability of a flight succeeding? First, in order to answer those, this study explored the literature standards in terms of aerospace reliability standards (especially in the rocket industry) and machine learning applied to imbalance dataset (rare events prediction) due to the IR of successful launches (chapter 3). This first part acted as the business understanding step of the CRISP-DM methodology which was detailed in the chapter 4. Secondly, the chapter 5 explained both data understanding and preparation phases through, respectively, the description and the distribution of the raw data, the cleaning and the feature engineering processes. After the data were processed, the models saw in the literature review could have been built and set up, all explained in the chapter 6. Once the models were ready, they were evaluated through a dedicated evaluation pipeline with different metrics which could fit the imbalance class problem, beyond the classical “accuracy” (chapter 7). The random forest algorithm seemed to show the best overall performance. Therefore, their results were interpreted in the chapter 8. SHAP gives a clear framework for those interpretations by ranking the features used by their impact on the launch status.