Conclusion

Thus we conclude these two months of master internship whose project is entitled "Management of missing data in building thermics" presented in this report with all the tasks we had to realize.

Unfortunately, we could not achieve all the objectives defined at the beginning of the project, in particular the comparative analysis of the performances of the two machine learning algorithms to fill the missing data of the Ibat project (Random Forest and Neural Networks), due to some constraints such as the coronavirus health crisis, which caused the projects to be delayed due to the recourse to teleworking which is difficult to manage for trainees.

The first step was to finalize the results obtained during the semester project, which consisted in carrying out the benchmark by making a random query of the nodes, as well as testing the performance of the imputation algorithm. The next step was to realize an LSTM neural network model that will allow to predict the temperature in the univariate and multi-variate case. The final step was to study the different comfort and discomfort indices in our data set.

This project has contributed very strongly to our training, notably with the acquisition of new skills and the mastery of new technologies that we used during its implementation, notably that of neural networks, more precisely LSTM (long short term memory), and the notion of comfort and discomfort indexes.