Grid Search for simple models for one population

[AgathaSchmidt]

Feature description

In the first implementations of NNs for one population, we randomly chose a model architecture and hyperparameters which by chance performed well.
Now, we want to analyze the three model types MLP, CNN and LSTM in more detail and find the optimal model architecture as well as the best hyperparameters. This way, we can identify the best model.
Additionally, some visualizations e.g. heatmaps should be implemented.

• grid search for secir simple
• grid search for secir groups