/kickstarter

Machine Learning Models to address Backers' and Creators' Concerns

Primary LanguageJupyter NotebookMIT LicenseMIT

Machine Learning Models to Address Kickstarter Backers' and Creators' Concerns

made-with-python MIT license

This is a Xccelerate Full Time Data Science Bootcamp Cohort 3 group project, that we practice our skills in building machine learning models.

Riding on the Kaggle dataset regarding kickstarter projects, we try to build ML models to address:

  • Creators' concern about the fund amount being raised
  • Backers' concern about whether a campaign be successful

Skills: data understanding, visualisation, data transformation, machine learning models

Table of Contents

For detail please refer to the jupyter notebook here.

  • Data Cleaning
  • Data Visualization
  • Regression model for Creators (Linear Regression, Random Forest Regressor)
    • Model 1: using predicting features that are only available before the campaign starts
    • Model 2: include features that are only available AFTER the campaign starts (explanation included)
      • Linear Regression
      • Random Forest Regressor
    • Summary of Regression problem
  • Classification model for Backers (Logistic Regression, Random Forest Classifier
    • Model: all are based on features available prior to campaign starts
    • Summary of Classification problem
  • Improvement

Presentation

Our presentation of the project can be found here. Google Slides