GSoC 2017 - Coding Period | Week 12

Throughout this week, I was mostly working on the sample Rails application along with Shekhar and also on the documentation of the daru-io gem repository. With the coding period coming closer and closer to an end, the focus is currently on turning the project into a deliverable open-source project, which could benefit from community (open-source) contribution.

• Repository : GitHub link
• Deployed Application : Heroku link

With the Rails application being deployed to Heroku, demonstrating the sample application is much easier. The Pull Requests Download as {format} and HTML Importer showcase have been merged and deployed. As it can be seen from the election controller, all the visualization entities like DataTable, Pie Chart and Column Chart have been achieved in just a couple of lines of code. Here's a code snippet of the controller -

  
  #! app/controllers/election_controller.rb

  class ElectionController < ApplicationController

    layout :resolve_layout

    def data
      @df_stock = Daru::DataFrame.from_html('http://eciresults.nic.in/PartyWiseResult.htm').first
      @df_stock.delete_row(-1) # delete the 'Total' row for visualization purpose
      
      data                = @df_stock.map_vectors { |x| [x.name, x.to_a.map { |y| y.to_i.to_s==y ? y.to_i : y } ] }.to_h
      new_df              = Daru::DataFrame.new('Party' => data['Party'], 'Won' => data['Won'])
      opts                = { adapter: :googlecharts, height: 500, width: 1000 }
      table_opts          = { adapter: :googlecharts, pageSize: 10, height: 300, width: 400 }
      @df_stock_table     = Daru::View::Table.new(@df_stock, table_opts) # DataTable entity
      @df_stock_pie_chart = Daru::View::Plot.new(new_df, type: :pie, **opts) # Pie Chart entity
      @df_stock_col_chart = Daru::View::Plot.new(new_df, type: :histogram, **opts) # Column Chart entity
    end

    private

    def resolve_layout
     case action_name
       when 'data'
        "googlecharts_layout"
       else
        Daru::View.plotting_library = :nyaplot
        "application"
       end
    end
  end
  

Great, the controller part is done. But for visualization, we obviously need to have the views part. Here's the most simple way to embed the DataTable or Charts.

  
  #! app/views/election/data.erb

  <%=raw @df_stock_table.div %>
  <%=raw @df_stock_pie_chart.div %>
  <%=raw @df_stock_col_chart.div %>
  

Here are a few screenshots showcasing the above example of election data with the use of HTML Importer.

Irrespective of how well written the code is, the software won't be friendly for users unless it is well documented. With daru-io to be transferred to SciRuby organzization on GitHub soon enough, it is quite necessary to add files : README, LICENSE, CODE OF CONDUCT, CONTRIBUTING and so on. They are being added quite exhaustively in this Pull Request, to ensure that users and contributors of daru-io, find it easier to develop with daru-io.

Meanwhile, a discussion regarding the calls of read-write methods from daru to daru-io is going on, in this issue. Also, Pjotr's suggestion of adding benchmarks for daru-io to identify the IO modules that might require faster approaches, sounds good and this is something that daru-io users could need, to decide on faster formats to import / export. Though this might not be possible within the GSoC 2017 timeline, I'd definitely like to add these benchmarks after GSoC 2017.