Simply upload a csv / excel / sqlite db with your desired data and select a target feature to predict.
The model will initially train itself and provide an RMSE value (better if close to 1 but depends on the size of data / number of features).
This RMSE number can then be fine tuned by specifying two hyperparametres:
- alpha which sets the capture net of the lasso model.
- iterations which train the model on for the specified number of epochs.
Once the model is fine-tuned, the ideal values can then be used to run on the desired dataset - it will automatically hunt for features in all the rest of columns and predict a value based on custon feature values.
- ML Cheatsheet 🔖
- SA Methods (including ISAMI guides) 📗
- Special thanks to CK for his amazing ideas 🧑 (future is Digital!)
Contributions are always welcome - please get in touch with me!
Goto goto/pioneer to pitch your automation ideas!
Please adhere to this project's code of conduct.
A simple search on google will yeild answers, but here - click me to learn more
Because:
- I wish to continue to develop to the point where I can 'deliver' E2E solutions quickly and efficently.
- Life is not worth living without challenges.
- It's a much nicer UX.
No. If you're looking to predict the RFs based on a given criteria for eg - this is meant to provide a general 'feeling' of what it should be, in some cases with upto 98% accuracy! While it will be capable of this in the next 5-7 years, the answer is sadly no today, keeping most of us employed.