Skip to content

Six-Kizuki-to-the-moon/Machine-Learning

Repository files navigation

Machine-Learning

This repository is used to create travel recommendation systems based on user preferences

Resource

  • Dataset : The first step we do is clean the dataset we got from Indonesia Tourism Destination that we will use and then we save it in the dataset folder in this repository.
  • Model : We creating the three models with using Tensorflow, Sklearn technology.
    • Content Based Filtering, for giving best preference based on user input such as category, city, and price. {model1, model2}
    • Collaborative filtering, for giving recomendations based on ratings user on every destination {model3}

About Member Machine Learning Path

Name Student ID Social Media
Mochammed Daffa El Ghifari M181DKX4080 LinkedIn
Dewa Putra Hernanda M131DSX0533 LinkedIn
Fadhila Salsabila M038DSY2929 LinkedIn

Documentation

  • API folder contains model Machine Learning & FLASK
    • For model machine learing on this folder we used mysql to read the dataset
  • dataset folder contains cleaned data for synchronizing the database from Cloud Computing
  • model folder contain model machine learning for recomendation system
  • file data_cleaning_visualization.ipynb for see the visualization results to get better understanding of the data that we used.

Model explanation

Recomendation Collab

This model focuses on rating predictions for users based on their ratings of other destinations and provides the closest tourist destinations by implementing the "haversine distance" method.

  • On the training process, we got the best graph with loss: 0.5242 - val_loss: 0.0807

    TrainValGraph!

  • Because we will train again this model on flask API with fewer epochs so we create pre-trained model for making faster computation & optimal result

    Type Ephocs time result
    Without pre-trained 200 2m,7.5s loss: 0.5242 - val_loss: 0.0807
    With pre-trained 10 3.4s loss: 0.9272 - val_loss: 0.4587
  • This model will be implemented on "Destination near you" section

    Implementation!

Recomendation Similar Item

This model focusing for giving similar item destination after user click or open the detail destination.

  • On this model we just using Mathematics like cosine similarity for getting similar item between place_name & place_category

  • This model will be implemented on "Another options for you"

    Implementation!

Recomendation Category

This model focusing to give the best destination from user preferences such as { Category, City, and Price } and after that will give the 3 options {Gold, Silver, Bronze} with 5 destinations on every options.

  • First, our model will filtering category, city, price, and rating from the database, based on user preferences

  • After that on this model we just using Mathematics like cosine similarity for getting similar item between price & rating

  • This model will be implemented on "Explore {Main Features}"

    Implementation! Implementation!

About

No description, website, or topics provided.

Resources

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages