Gracefully Handle Missing Data

[sjtrny]

Some series will have missing data (NaN) at random time points. We need to come up with a mechanism for handling this.

Some ideas that spring to mind:

• carry forward last good value
• carry backward the next good value
• remove all time points before a missing value (i.e. only use most recent string of good data)

Example source

{"title":"Inflation for Rwanda (World Bank)",
  "source":"WorldBank",
  "url":"https://api.worldbank.org/v2/country/RWA/indicator/FP.CPI.TOTL.ZG",
  "frequency" : "AS",
  "tags" : ["Rwanda", "Inflation", "Economic", "Africa", "World Bank" ]
  },