Skip to content

Code to analyze diverse data and predict wind velocity with tensorflow. It employs a neural network architecture to train and optimize the model for regression problems.

License

Notifications You must be signed in to change notification settings

FernandoSchett/wind_speed_prediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🌬️ Wind Speed Predtion Model 🌬️

Developed by 💻:

Special thanks to 🥰:

  • Murilo Boratto, my advisor for their invaluable guidance throughout this project.
  • Anúsio Correia, your experience, knowledge and reference material have been invaluable to my progress.

About 🤔:

The project aims to solve a regression problem by predicting wind speed based on a given dataset using the TensorFlow library. The code was developed as part of an internship project and is available at this link. However, towards the end of the project, a version in PyTorch was used, so the repository was created specifically to contain the TensorFlow version.

The problem statement revolves around the significance of wind speed in determining the growth and yield of oranges. The farmer has gathered historical data on wind speed and its suitability for orange cultivation. Each sample in the dataset consists of a measurement of wind speed in meters per second, available as wind_data_train.csv in the datasets folder.

The objective is to develop a prediction system that can determine wind speed based on the provided historical data. To achieve this, the model needs to be trained using the given data and then tested using a separate test dataset named wind_data_test.csv, which can be found in the datasets directory.

Resourses 🧑‍🔬:

  • Wind Speed Prediction: The project is capable of predicting wind speed based on the provided dataset.

  • Training and Testing: The model is trained using the historical data and tested using a separate test dataset.

  • Data Preprocessing: The project includes data preprocessing steps such as normalization, filtering out unnecessary data, and handling missing values (NaN) in the CSV file.

  • Performance Evaluation: The project provides comparative graphs between the expected and predicted wind speeds, allowing for an assessment of model performance.

  • Loss Monitoring: The project tracks the loss of the model during training and provides a graph showing the loss over epochs, enabling analysis of model convergence and performance.

  • Comprehensive Data Visualization: The project includes a graph that displays all the collected data, providing a visual representation of the dataset used for training and testing the model.

Results 📈:

Figure 1 - Collect Data Graph Comparison.

Figure 2 - Model Loss X Epochs Trained.

Figure 3 - Final Result Graph - Predicted(Orange) X Expected(blue).

Dependencies 🚚:

In summary, heres what you're gonna need in order to run the project:

  • python3
  • tensorflow 2.12.*.
  • numpy 1.23.5.
  • pandas 2.0.2.
  • matplotlib 3.7.1.
  • pydot 1.4.2.
  • graphviz 2.43.0.
  • keras 2.12.0.

For installing dependencies more quickly, you can run the following command at terminal, inside the clonned repository:

pip3 install -r dependencies/requirements.txt

How to run it 🏃:

First, clone this repository. After that, simply execute the wind_speed_prediction.py file with the command:

python3 src/wind_speed_prediction.py

Development Process ⚙️:

This project was developed without following a specific methodology, allowing for flexible work hours. Feedback from mentors was received once a week to guide the development process.

Tools Used 🛠️:

How to contribute 🫂:

Feel free to create a new branch, fork the project, create new issue or make a pull request contact one of us to develop at this code.

Licence 📜:

Apache V2

About

Code to analyze diverse data and predict wind velocity with tensorflow. It employs a neural network architecture to train and optimize the model for regression problems.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages