hannah-ross / dataforgood

python data science project

dataforgood

This is Team Campus Cults' final project analysis of Kickstarter data for Statistics 131 at UCLA.

Our final report is in the jupyter notebook "Final_Report.ipynb"

A supplemental exploratory analysis notebook "Plotly_Supplemental_Exploratory_Analysis.ipynb" can be viewed here: http://nbviewer.jupyter.org/github/hannah-ross/dataforgood/blob/master/Plotly_Supplemental_Exploratory_Analysis.ipynb

Our project presentation video can be viewed on youtube here: https://www.youtube.com/watch?v=gSZCPRbuOD8&feature=youtu.be

Our presentation slides themselves can be viewed in "project_slides.pdf"

Project Organization

data

a folder containing the Kickstarter data set

background_info

a folder to document background information, common knowledge, and discussion of variables

• discussion of each of the variables included in the dataset, and what the observational unit is
• discussion of how the data is collected
• “Common knowledge” shared among relevant parties involved
• a review of other related research or studies that have been done

exploratory_analysis

a folder for code exploring the data (view in http://nbviewer.jupyter.org/ )

• any data cleaning and new derived features
• summary statistics
• outliers
• interesting relationships
• readable tables, graphs, and explanatory commentary

modeling

a folder for code fitting a statistical model to data for prediction

• explain decisions regarding the choice of model, and the reasoning behind the inclusion of predictive features. (Probably backed up by the findings in the exploratory data analysis.)
• seperate training and testing data to fit a model for predictive purposes
• cross-validation to ensure that the model is not overfitting features unique to the training data. A metric will need to be selected to show the predictive performance of the model.