Kiva CrowdFunding

Storytelling with Interactive Data Visualization


Approach:

Following the Hybrid Model of Data visualization with balancing Autor-Driven and giving flexibility to the Reader driving towards Reader-Driven approaches in each slide.
Slides with Charts starts with Interactive slideshow and then user can drilldown with various parameters, triggers.

Rules:

Horizontal Scrolling:

Follows Reader-Driven and Drilldown approach ( expands the informations shown on the main slide but return back to current slide parent),
Reader can navigate to any of slides in that scope.
In each Drill-down slide reader was allowed to use triggers to narrow data selection through either setting data filters to specific dictionary values or/and achieve similar result by visually choosing regions.
  (e.g. rectangles) on interactive charts, so only values associated with regions will stay in data selection.
Visual Structuring: user was provided dots at the bottom of the screen, enables to know where they are and giving them flexibility to jump as needed.

Vertical Scrolling:

Follows Interactive Slideshow. Allowing user to navigate through Author-Driven approach.


Slide Horizontally for Drilldown.
Slide Vertically for Interactive Slide show.

Click right Arrow now ->

Overview

Data Source:Kiva CrowdFunding

Kiva.org is an online crowdfunding platform to extend financial services to poor and financially excluded people around the world. Kiva lenders have provided over $1 billion dollars in loans to over 2 million people.

Prepare Data

  1. Upon data files downloaded from Kiva.org
  2. Data was cleaned using Data Preparation tools - Tableau Prep
  3. Corrected for missing values and populated with 0's as needed.
  4. Aggregation was done into separate files for easy and flexibility for better visualization in d3.js
Scroll down horizontally to take to next section or click on right arrow to take to first slide(home).

Essay

Overview:

Upon partial analysis of the data, trying to showcase available datapoints into Data visualization for better understanding the crowdfunding with respect to timelines, sectors, country, partners etc.
In this effort, i was able to concentrate on timelines, sectors, countries, partners and ignored many other data points that are available in the dataset.
Below are the two approaches built to better convey the understanding.

Narrative:

Following is the process following in storytelling or Narrative Visualization
Visual Structuring

  1. Genre -> Incorporating the Hybrid Model (Author-Driven and then Reader-Driven) for Narration.
       Author-Driven: Vertical Scrolling enables Drill down model in between scenes.
       Reader-Driven: Horizontal scrolling signifies Interactive Slideshow in traversing between Scenes and
  2. Design Dimensions:
       Highlighting: Each slide has annotations to focus on important events
       Tranistion guidance: Slides with drill down has left and right arrows
       Messaging: Each chart enabled with Tooltips to inform the required message.
       Visual Structuring: Enabled scrollbar for vertical navigation(Author-Driven) and dots ( bottom of screen) for Horizontal slides(Reader-Driven)
  3. Two different Scenes are focussed. Timeline based on Funding Amount and Country Based.
  4. Funding Amount Displayed in terms of Millions with 2 digits prefix.
Move to next slide for detailed information on Scenes.
Note: Using FullPage.js and you will see some Licensing errors if watched in Developer Tools. Please Ignore and it won't affect the usage.

Charts:

  1. Timeline Based:

    All charts in this category are based out of Funding(continious) dispersed considering various attributes.
    ex: Total Funding in each month, Total Funding across sectors.
    • 1.1: Total Funding Projected by Month - Simple Area Chart
      x-axes: Month-Year(Continious) (range: date extent),
      y-axes: Funded (Continious in Millions) [ Sum of funding across all sectors]( Range: 0 to max(funded + 1M )). Added 1M to extend y-axes for better visibility.
      In trying to consider funding growth (consolidated over individual data values) is better displayed in Area Chart. This chart gives overall projection of spending over time.
      Equipped with Line graph for better projection of Total funding in each defined month.
      Created Annotations during peak dates with Month-Year and Funded Amount in Millions.
      Provided tooltip for projected line circle in Millions
      All scales are Linear.
    • 1.2: Funding By Sector - Stacked Area Chart
      x-axes: Month-Year(Continious)(range: date extent),
      y-axes: Funded (Continious in Millions)
      Following previous chart, this chart will dig deeper into above scene and display in terms of Involved Sectors. Since we are consolidating Individual Sectors is better displayed using Stacked Area Chart.
      Projected Annotations for the peak Year-Month and Funded amount in Millions.
      Highlitghted Highest and Lowest Funded Sector.
      Displaying name of sectors side-by-side for better display and avoid overlapping.
      All scales are Linear.
    • 1.3: Sector Funding(Filters,Parametrized) - Bar Chart
      x-axes: Sectors(Discrete)(range: sectors Ordinal),
      y-axes: Funded (Continious in Millions) ( range: 0 to max+1e6)
      Used schemeCategory20c color scheme.
      Following above, Ideally, displaying Bar chart makes more sense as we have definite Sectors.
      This chart displays a visual way to identify top and bottom funded sectors.
      Highest and Lowest funded sectors are displayed and Readers can filter based on Sectors.
      All scales are Linear.

  2. Country Based:

    All charts in this Scene are based out of Country(discrete) and spent funding till now.
    • 2.1: Total Funding By Country
      x-axes: Country(Discrete)(range: country Ordinal) ,
      y-axes:Funded Amount(Continious in Millions) ( range: 0 to max).
      Considering the values of Discrete and Continious, Bar chart if very appropriate. This simple Bar chart is to project Funding between countries.
      Provided appropriate Sorting Triggers to get sense of spent. Providing Sort by Funding and Country gives more idea on Funding distribution.
      Annotations, Legends are displayed appropriately.
      All scales are Linear.
    • 2.2: Partner/Sector Distribution By Country - Stacked Bar Chart
      x-axes: Country(Discrete) (range: country Ordinal),
      y-axes:Partners and Sectors Count.
      Following above scene, user can drill down and get more info on Partners and Sectors Count in each Country.
      Use different filters, Triggers to get to understand with various combinations.
      Annotations, Legends are displayed appropriately. Annotations are not displayed if search has that specfic country outof scope.
      All scales are Linear.

Theme: Timeline Based Funding

Area Chart - Total Funding

Below chart displays the amount of funding dispersed by Kiva CrowdFunding in each of the month starting from 2014-Jan to 2017-July Hover the chart to see the funding in each projected month. ( Note: There is a drop in July 2017 has there is no much data available or missing in the dataset)
Note: Date can be discrete and continious based on situation.
  • Funding (continious) is taken as y-axes and other attributes( ex: Sectors [discrete] are taken as x-axes)
  • Three Annotations are created for the Peak Month ( March of each year)
  • Scatter plot was drawn for easy identifying each month on the area chart.
  • Mouse hovering will provide details on Funding in Millions.

Annotations: Noticed that there is typical spike in March in each year except 2014. It might be because of typical year end to most of the countries.

Stacked Area Chart - Sector Funding

  • Continuing the prior slide, this slide is expanded to show the funding dispersed in various sectors.
  • After reviewing the above chart, its good idea to drill down and see which Sector is getting more funding.
  • This Stacked Area Chart is drawn considering these factors into account.
  • Annotations are created for the same Peaks but main intention is to get the size of allocations.
  • By viewing it can be figured out that Agriculture, Food and Retail takes more share. But, again its not completely clear in understanding the sequence of sectors allocations. This is being addressed in next slide.
  • Mouse hovering displays info on hovering sectors, complete total funding.

    • Legend schemeCategory20c for coloring Sectors

    Annotations: Spikes in March is noticed espically from 2015 onwards.

    Bar Chart - Sector Funding(Filters,Parametrized)

  • Continuing the prior slide, this graph helps to view the content filtering parameterized selection of Sectors.
  • Here is the bar chart which gives more detailed info for digesting the funding across sectors and their ranking.
  • Sectors are ordered in Highest to lowest and its more evident that Agriculture takes the first rank and Wholesale takes least funding.
  • Annotations are created for the same.
  • Apart from that, this slide is also equipped with Parameterized searching using Checkboxes and Refresh button.

    • Legend: schemeCategory20c for coloring Sectors

    Annotations: Highest Funding and Lower Funding Sector


    Theme: Country Based Funding

    Bar Chart: Total Funding

    Below chart displays the Funding dispersed for different countries. x-axes takes countries and y-axes takes the continous variable Funding amount.
    This gives idea of 1st and 2nd top Funded Country. Users can sort the data based out of country or funded amount using radio buttons.
    Annotations: Phillipines taking 1st top funded country followed by Kenya.

     Sort by Funded  Sort by Country

    Legend

    Stacked Bar Chart - Partner/Sector Distribution

    • Once reviewing about chart, reviewing the sectors, partners in each of these countries would give us more info and we can even see if there is any co-relation.
    • Drawing a Stacked Bar Chart with Countries and Stack Counts of Partners and Sum of Sectors.
    • Here we have used annotation for Highest Partners (Kenya).
    • Provided a way to Sort based on Total and Revere Partners and Sectors to get some insights.
    • Provided a third Filter of filtering Countries who had more than 5 Partners( Avg partner count)

    Annotations: Kenya takes the highest in Partners with ratio of 3 partners per sector.
    Use Sort and Reverse checkboxes to get sense of partners and sectors for each country.

     Sort by Total Count  Reverse Layout  Average Partner Countries(> 5)


    Conclusion

    Methodology

    Following the Hybrid Model (Author-Driven and Reader-Driven) using Interactive Slideshow
    Charts are Based on:
    1. Timeline Based: Main strategy around these charts is to find the funding spread across different sectors with various timelines into scope.
    2. Country Based: Main strategy is to find the funding spent in each country and what other factors are being contributed like Partners, Sectors etc.

    Conclusion:

    1. Considering the Analysis and this Visualizations ,found that there is pretty good distribution of Sectors across different countries but there is lot variation on Funding.
    2. More Partners, more Funding
    3. Funding spikes during month of March and sometimes December.
    4. Can be Expanded:
      a. Partners specific Performance
      b. MPI Data analysis ( data available in kiva website)
    Github Source Code
    Thanks for Reviewing!
    Click Home