To start off, clone this branch of the repo into your local:
git clone -b main --single-branch https://github.com/aL0NEW0LF/data-playground-desktop
After cloning the project, activate the env:
.venv\Scripts\activate
You can run the following command to install the dependencies:
pip3 install -r requirements.txt
Then run the main file with:
python main.py
2 sets of data are included, for users to test the app, first one is the titanic dataset to test all features of the app, second one is the same dataset but cleaned so the user can test training and testing the model directly.
Important
The work flow will be as follows:
After starting the app:
- Choose your wanted ML algorithm
- Upload data
- Choose your target column
- Proccess & visualize your data
- Split your dataset into training & testing data
- Train & test your model, then see results
Note
- Choose target column right after uploading data
- Variance threshold feature selection
- More data processing
- More plots if possible
- Better error handling & conditions handling
- Add training & testing the model without spliting the data
- Model configuration
- Model, metrics & data visualization after training
- Better UI
- Import ML model to use
- Add log file to record every step of the process and be able to return
- Fix the 'K-means' workflow, with more visualization of the results
- Add 3d and violent visualization
- Choose what form you want to save your file
- Add more regression models
- Add regression metrics plots
- Add a button to show the dataset info and description and different metrics on click
- Add a way predict for new datasets
- Add a way to create a new dataset or line of data
- Save the label encoder used to later reverse transform future predictions
- Choose the column where you want to handle missing values