Features and ML model selection - Githubissues

TechLabs-Berlin / st21-beachblue

The idea behind Beachblue is to help everyone discover their favourite beach from the comfort of their own home. Planning for beach days made easy, or as we say: Your next dip, right in one click!

2 stars 2 forks source link

Features and ML model selection #6

Open Nidhi-1211 opened 3 years ago

Nidhi-1211 commented 3 years ago

List of features to be considered for ML model
https://github.com/TechLabs-Berlin/st21-beachblue/blob/develop/cols_to_keep.xlsx

Nidhi-1211 commented 3 years ago

Testing using KModes
reference links
https://shapeofdata.wordpress.com/2014/03/04/k-modes/
https://www.kaggle.com/ashydv/bank-customer-clustering-k-modes-clustering

Nidhi-1211 commented 3 years ago

-Code for Kmodes

https://github.com/TechLabs-Berlin/st21-beachblue/blob/recommender-system/playas-kmodes.ipynb

Nidhi-1211 commented 3 years ago

Kmodes model didn't give results as expected

Nidhi-1211 commented 3 years ago

Idea -
to use google ratings to predict ratings for spots having similar features but no ratings
to use features assign them with weights and categorize them in the range of 1-5.