The purpose of Build Week is to empower students to demonstrate mastery of the learning objectives. The Build Weeks experience helps prepare students for the job market.
The Pitch:
Using NLP and / or regression techniques, Kickstarter Success can help predict how successful a kickstarter campaign will be based on the monetary goal, description, campaign length, or catagories.
MVP:
- Train a model that predicts
campaignsuccess or failure (binary target variable.) - Deploy a model via Flask API so that predictions can be displayed to the
user.
Heroku App:
https://kickstart-campaign-prediction.herokuapp.com/
Setup Details:
Required Packages:
numpy, python, dash, requests, pandas, scikit-learn, joblib, gunicorn
To start locally in command line:
clone this repository
start venv
command line code: python usd_app.py
Datasets Used:
The KNN Nearest Neighbors model was trained on Kickstarter campaign data from Kaggle:
https://www.kaggle.com/kemical/kickstarter-projects?select=ks-projects-201801.csv
API Used:
Currency exhange rate API used to convert all currency inputs to USD (key required):
https://www.exchangerate-api.com/
Meet the Team:
Nicholas Papenburg; Github: https://github.com/NPAPENBURG
Celina Walkowicz; Github: https://github.com/CelinaWalkowicz
Matt Grohnke; Github: https://github.com/mgrohnke