Mathematics
2024, 12, 571
15 of 26
4.6. Implementation Details
In this study, the implementation of the ensemble model was skillfully executed using
Python for its wide-ranging library support, particularly TensorFlow and Keras for DL,
alongside Scikit-learn for data preprocessing. Pandas and NumPy complement these for
effective data manipulation. Version control was meticulously managed using Git, with
the project’s codebase and version history accessible at the repository
https://github.com/
TATU-hacker/CNN-LSTM-GRU.git
, uploaded on 17 November 2023. The computational
backbone of the project was the Kaggle GPU P100 platform, known for its formidable
processing capabilities, which significantly expedited the training and inference phases.
To address the constraints of IoT environments, the ensemble model was designed
with scalability and efficiency at its core. It can adapt seamlessly to varying data volumes,
a critical feature for IoT applications. To ensure compatibility with IoT devices, known
for their limited processing capabilities, the model was optimized for computational and
memory efficiency and tailored for potential integration with edge computing, thereby
minimizing latency and reducing bandwidth requirements. This thoughtful combination
of software choices and hardware optimization ensures the model’s applicability in the
dynamic and resource-constrained landscape of IoT-based EVCS.