Evaluation

[bensoltoff]

Evaluation of final project by Benjamin Soltoff

Topic	Excellent	Satisfactory	Needs Work
Coding style		x
Coding strategy		x
Presentation: graphs	x
Presentation: tables		x
Achievement, creativity		x
Ease of access		x

Remarks:

• Figure for time banks by state - would look much better graphed spatially using a map. Could look at ggmap or building a chloreopath
• The shiny app is nicely integrated into the site - adds value by allowing users to search through the timebanks
• For the correlation graphs, look into integrating correlation statistics to quantitatively confirm your visual interpretations
• Why dichotomize the activity level for the decision tree? You can use decision trees for regression problems (i.e. continuous variables). It feels like you throw away good data by converting to a binary variable
• I'm not sure I follow your interpretations of the decision tree variables. It directly conflicts with the variable importance plot you generated, which shows that the total number of active members is the most important variable in the decision tree. Removing that variable greatly increases the error rate
• This is great start for a research paper, but you will need to develop a better understanding of the statistical model to accurately interpret the results
• Your code is generally clean and well-documented
• Organization of the repo could be improved somewhat. You've split it into the pipeline with all the important code and functions, then generate the website from the graphs previously generated. Clean it up more with a descriptive readme file to describe the repo and how someone should run the files to reproduce your findings