3.3. Evaluation of Model Results
The classification model is primarily carried out by separately training
3 different datasets from the locations. In the second step, a single
dataset was obtained as a result of combining all datasets. In the
developed LSTM model, this dataset was also trained and the results were
obtained.
First of all, the training of three different locations was carried out
separately and the results are shown in Figure 8. The model was
completed with 200 training rounds, but the accuracy values were
provided in a maximum of 100 training rounds, and then the model
accuracy was fixed. As a result of training the data from the locations,
respectively, 1st location 85%, 2nd location 84% and 3rd location 84%
accuracy value. As a result, the fact that the temperature and humidity
values are in the data set that he learned kept the order
classification accuracies high. Afterwards, all datasets were combined
and the training of the model was performed again and 94% accuracy was
obtained. The most important reason for the increase in the accuracy of
the model is due to the significant increase in the amount of data in
the data set.