Résumé:
With the rapid development of the so-called cloud storage. The practice of storing and
sharing scientific experimental data on different internet platforms grew tremendously, which led
to a significant accumulation of data. Until recently, this large quantity of data has been neglected
due to the lack of effective techniques to gather knowledge and useful information from these data.
Nevertheless, the progress achieved in the data-driven techniques in the past decade offered unique
opportunities to extract important information from the material data. In this thesis, we have used
advanced techniques of machine learning to solve critical problems that prevent successful
commercialization of the perovskite solar cells technology through the investigation and the
analysis of an important amount of data consisting of the measurements and materials information
related to the manufacturing and the operation of these devices. This research provides a practical
and useful guide for improving the performance of this kind of solar cell as well as enhancing the
operational lifetime. Moreover, we have compared the results of machine learning with different
previous experimental research, where we found remarkable coincidences. Which opens up
important prospects in this field.
