Kickstarter Success is a machine learning powered application to help you predict whether your kickstarter campaign will be successful based on several factors: campaign name, description, category, fundraising goal, country, and duration.
The data science team:
- Used Kickstarter dataset to predict the percentage of success of a campaign using TfidfVectorizer and Spacy models to do natural language processing on campaign descriptions.
- Created and deployed Random Forest models using a Flask API on Heroku so the back end can make requests and return predictions.
You can check out the project at https://kickstart-success.netlify.com
Role: Machine Learning Engineer
DS Tech Stack: Python, SQL, Flask, Heroku, Scikit, Spacy, TfidfVectorizer, Plotly, AWS (Sagemaker, S3)
Timeline: 1 week
The GitHub repo with all of the project notebooks and code can be found here.