- Get the latest evolution.ipynb (
git pull origin evolution
). - Upload evolution.ipynb to your working Google Drive folder, rename for different input options if needed.
- Open evolution.ipynb with Google Colaboratory.
- Run the setup Google Colaboratory cell to set up the folder structure in Google Colaboratory’s VM.
- Upload model.py and dnn_regression.py in models to the models folder.
- Upload all required data CSV to data/stock_prices folder.
- Upload build_dataset.py and options.py.
- Run all import dependencies cells.
- Change
input_options
in initialize models section to working input options (refer to input options document). - Change
stock_code
in initialize models section to working stock code. - Change
ITERATIONS
in Evolution algorithm section based on available time. - Run all cells in “Get last run data”, “Initialize models”, “Initialize errors”, “Evolution algorithm” and “Write this run data” sections.
- last_run.json, evolution_tensorboard_logs.zip and evolution_model_graphs.zip will be generated after running cells in “Write this run data” section.
- If time allows, re-run all cells, starting from “Get last run data” section, for another run of
ITERATIONS
evolution iterations. New last_run.json and 2 zips will be generated, with data from all previous runs. - Download last_run.json, evolution_tensorboard_logs.zip and evolution_model_graphs.zip for next time.
- Repeat all steps, with an extra step of uploading last_run.json next time.
- last_run.json will be appended with new data, download and replace the old file. evolution_tensorboard_logs.zip and evolution_model_graphs.zip will have data only from a particular Google Colaboratory session, download them with another name and manually merge the content.
- Upload last_run.json, merged evolution_tensorboard_logs.zip and merged evolution_model_graphs.zip to Google Drive.
- Run all cells in “Plot predictions” and “Plot evolution data” section.
- Repeat all steps for another input options.