Gradual improvement on a road to 1.0

We are currently working hard on a proper version 1.0 release, with daru-io integration, full codebase cleanup and a lot of cool things.

In the meantime, here is 0.2.0!

• Major Enhancements

  • Add DataFrame#which query DSL (experimental! @rainchen)
  • Add DataFrame/Vector#rolling_fillna (@baarkerlounger)
  • Add GroupBy#aggregate (@Shekharrajak)
  • Add DataFrame#uniq (@baarkerlounger)

• Minor Enhancements

  • Allow Vector#count to be called without param for category type Vector (@rainchen)
  • Add option to DataFrame#vector_sum to skip nils (@parthm)
  • Add installation instructions to README.md (@koishimasato)
  • Add release policy documentation (@baarkerlounger)
  • Set index as DataFrame's default x axis for nyaplot (@matugm)

• Fixes

  • Fix DataFrame/Vector#to_s when name is a symbol (@baarkerlounger)
  • Force Vector#proportions to return float (@rainchen)
  • DataFrame#new creates empty DataFrame when given empty hash (@parthm)
  • Remove unnecessary backports dependencies (@zverok)
  • Specify minimum packable dependency (@zverok)
  • Preserve key/column order when creating DataFrame from hash (@baarkerlounger)
  • Fix DataFrame#add_row for DF with multi-index (@zverok)
  • Fix Vector#min, #max, #index_of_min, #index_of_max` (0.1.6 regression) (@athityakumar)
  • Integrate yard-junk into CI (@rohitner)
  • Remove Travis spec restriction (@zverok)
  • Fix tuple sorting for DataFrames with nils (@baarkerlounger)
  • Fix merge on index dropping default index (@rohitner)