cvanoort/datalab_challenge_2

Data Lab (CSYS 395, Spring 2020) Challenge #2

• Assigned 2020/02/18
• Due 2020/03/10
• Team 3:
  • Anoob Prakash
  • Jessica Cole
  • Elizabeth Espinosa
  • Erik Brown
  • Samuel Rosenblatt
  • Colin Van Oort

Problems

1. Parse the SMAC data files.
2. Reproduce the analyses from the SMAC paper.
   • Number of Community Visits time series
   • Burial by Chiefdom and burial type time series
   • Community By-laws time series
3. Identify surprising chiefdoms: Regions that had a lot more or a lot less cases than neighbors. Do they stand-out in social mobilization data?
4. Continue the exploration in interesting directions:
   • Compare the attack rate in people to the attack rate over chiefdoms.
   • Is there any evidence that misinformation or distrust in the intervention lead to lower rates of reporting, safe burials and referrals?
   • Are social mobilizers from different organizations getting similar ratings in the different evaluation metrics?

Solutions

Resources

• Dataset
• Paper 1
• Paper 2

Notes on Sierra Leone:

• Divided into 5 administrative regions
• Further divided into 16 Districts
• Further divided into 186 chiefdoms
• Chiefdoms may be cut into "Sections" (ref 1, ref2)
• District maps can be found here

Repo Notes:

• When looking at the column discrepancies that are output by etl.py, it can be useful to combine multiple related discrepancy files with the following command:

  cat *.json | sed 's/,//g' | grep -v -E '\[|\]' | sort | uniq 

  In particular, you can reduce the globbing to a category of interest, such as cat *Chiefdoms.json ....