Kickstarter a crowdfunding website helped people raise a total of $1.6 Billion so far. For so many campaigns that succeed on kickstarter, there are are even so many that do not. My goal is to create a predictor website that helps fund seekers to predict whether their campaign will suceed or not.
- Get Kickstarter campaigns from 2013-2019 from Kickstarter scraper website
- Download data set into MongoDB and do some data preprocessing
- Load MongoDb Data into Pandas
- Create some EDA and Data visualization
- Best times and days to create a campaign that will suceed?
- What categories of campaigns get funded the most?
- Connect that information to US market for those categories to see whether the size of market affects how they are funded ?
- Connect that information to US market for those categories to see whether the size of market affects how they are funded ?
- Best times and days to create a campaign that will suceed?
- Feature Engineering and selection of Kickstarter campaign observations
- Utilize classification algorithms such as Neural Networks, Logistic regression for prediction
- Perform Natural Language Processing on some of the columns e.g the text description of the campaigns is suitable for sentinment analysis and topic modeling .
- Build predictor output website
- Kickstarter
- USA market data
- MVP: Perform EDA visualizations, Feature selections using classification algorithms and Topic Modeling to extract topics on descriptions.
- Final: Build boostrap designed Flask predictor Website to allow people
- Selecting the proper features to predict the result of a campaign might be challenging as my datasets contains about 25 different columns .