Fish Finder
- 2 Devlogs
- 33 Total hours
A comprehensive expert system companion codebase that provides the heavy analytics as part of the analysis and prediction.
Wowzers! I have a website presentation now!
Pretty much what I spent alot of my time on was actually creating a HTML Website page presentation for the findings/creative processes that went into producing this system.
Since the last log, I have done a few things:
- Re-run the neural network training script, leading to an increased F1 probability score (60.1% -> 64.3%) which is very good!
- Created a HTML website with Firebase Web App Hosting services (Accessible at: https://fishfinder-presentation.web.app/).
- Conducted source file cleaning to remove further invalid records from the data source, reducing inaccuracies in predictions and models.
- Got the entire offline capabilities working (Tested by taking my MacBook outside to the local marina without WiFi, Cellular, or any otherwise connections to external networks.
- Fixed up some UI issues/bugs with the StreamLit application.
Sooooo, this project is beginning to near its end! Thank you for reading, and I hope that at least one person reading this may be able to actually use this themselves, or introduce this to someone else to use for their benefit!
P.S: If anyone has further questions or inquiries about how this works, or how to get it working on different types of devices, just drop a comment down below and I will reply promptly either directly through this website, or we can contact through Slack!).
Ta!
Wow, more work done!
What has been done:
- StreamLit website working, read the README.md file for instructions on how to host locally.
- Presentation website created: FishFinder-Presentation.web.app!
- Neural Network trained on ~40,000 records from within the dataset (20/80 split of testing and training).
- Dataset enrichment from the Open-Meteo Historical Weather API (with permission), introducing columns to the dataset for: Amount of Rain, Wind Direction, Moon Phase (calculated separately through a Python script), and Sea Surface Temperatures (returned “null”, so removed).
- Improved on PowerBI dashboards and Excel Pivot Tables and Charts.
Keep an eye out for more updates!
