Difference between revisions of "Apple Crunch: Proposal"

From Visual Analytics for Business Intelligence
Jump to navigation Jump to search
 
(8 intermediate revisions by 2 users not shown)
Line 25: Line 25:
  
 
<!-- Start of Problem & Motivation -->
 
<!-- Start of Problem & Motivation -->
== Proposal ==
+
[[File:Kick Finisher Logo 2.png|800px|frameless|center]]
 +
== '''Proposal''' ==
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
! Problem and Motivation
+
! '''PROBLEM & MOTIVATION'''
 
|-
 
|-
 
|}
 
|}
Line 38: Line 39:
 
There is a need to understand what makes a successful campaign. Are there any attributes that sets apart projects that are successful? The team would like to create several visualizations on kickstarter campaign data to analyze trends and determine if certain attributes of a campaign such as "amount of funding set" affects its chances of succeeding. In the end, the team hopes that others can use the visualizations created in order to have a better understanding of how they should create their campaign to have a higher chance of success.
 
There is a need to understand what makes a successful campaign. Are there any attributes that sets apart projects that are successful? The team would like to create several visualizations on kickstarter campaign data to analyze trends and determine if certain attributes of a campaign such as "amount of funding set" affects its chances of succeeding. In the end, the team hopes that others can use the visualizations created in order to have a better understanding of how they should create their campaign to have a higher chance of success.
 
<!-- End of Problem & Motivation -->
 
<!-- End of Problem & Motivation -->
 
  
 
<br>
 
<br>
Line 45: Line 45:
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
! Data
+
! '''DATA'''
 
|-
 
|-
 
|}
 
|}
Line 91: Line 91:
 
|}
 
|}
 
<!-- End of Data -->
 
<!-- End of Data -->
 +
 +
<br>
 +
 +
<!-- Start of Charts -->
 +
{| class="wikitable"
 +
|-
 +
! '''VISUALISATION DASHBOARD & CHARTS'''
 +
|-
 +
|}
 +
{| class="wikitable"
 +
|-
 +
! Chart !! Description and Analysis
 +
|-
 +
|
 +
<center>'''Visualisation Dashboard'''</center>
 +
 +
|-
 +
 +
| [[ File: 1.01.png|thumb|350px|center]]
 +
<br>
 +
<center> <b> Drill down Tree-map : Main Category </b> </br>
 +
 +
||
 +
The first visualization that we are going with is the “Drill-down Tree-map” with the size being the number of campaigns and the color being the category. The visualization showcases all the Main-Categories and Sub-Categories with their respective size being the number of campaigns. The main category is stated at the top left corner with the number while the sub-categories are faded in the background. The visualization contains the data of campaigns between 2015 and 2018 – which indicates the recent trends on the Kickstarter platform.
 +
 +
|-
 +
 +
| [[ File:1.02.png | thumb |350px|center]]
 +
<br>
 +
<center><b> Drill down Tree-map : Sub Category </b></br>
 +
 +
||
 +
This example visualization showcases the number of campaigns for each of the sub-categories of the “Film & Video”. From here we can see that under “Film & Video”, “Short Film” and “Documentary” are the most popular. We can interact with the tree-map and have the same exploration with other main categories as well such as “Music”, “Publishing”, “Theater” …
 +
 +
|-
 +
 +
| [[ File: 1.03.png|thumb |350px|center]]
 +
<br>
 +
<center><b> Stacked Bar Chart </b></br>
 +
 +
||
 +
In the stacked bar graph, we use the data across all categories from the year 2015 -2018. We calculate the percentage of successful projects and failed projects for each of the state. The states are then sorted based on the percentage of success in a descending order. From the visualization above we can see that the top 5 states with the highest percentage of successful projects are “New York”, “Rhode Island”, “Vermont”, “Massachusetts” and “Montana”.
 +
 +
|-
 +
 +
 +
| [[ File:1.04.png|thumb |350px|center]]
 +
[[File:1.05.png|thumb| 350px|center]]
 +
<br>
 +
<center><b> Interactive Visualization </b></br>
 +
 +
||
 +
The interactivity of the visualization comes in when we interact with the tree-map. After every selection of main- category or sub-category of the tree-map, the bar-graph visualization would be filtered based on the selection.
 +
 +
From the visualization above we can see that even though across all categories, “New York” has the highest percentage of successful projects, however, if we zoom into “Publishing” category, “West Virginia” is a better place to launch the campaign as compared to “California” as it has a higher percentage of successful projects specifically for the “Publishing” category. The above is just one example of the interactivity of the visualization. The user would be able to do the same thing and view the top 5 states with the highest percentage of success for any main-category or sub-category that they select on the tree-map.
 +
|-
 +
 +
| [[ File:1.14.png|thumb|350px|center]]
 +
<br>
 +
<center><b>  </b></br>
 +
 +
||
 +
From this visualization, we can see that “Dance”, “Theater” and “Music” have the highest percentage of success if we look at the overall data across all the states. However, if the user selects “Minnesota” as the state that they are interested in, they would get the following:
 +
 +
|-
 +
|-
 +
 +
| [[File:1.15.png|thumb|350px|center]]
 +
<br>
 +
<center><b> </b></br>
 +
 +
||
 +
From this, we can see that unlike other states, Minnesota is more popular with “Music” and “Comics”. From the user perspective who are thinking about their project idea to launch in Minnesota, a possible insight from this is that they should take a look at the category of “Music” and “Comics” since these two have relatively high percentage of success.
 +
 +
|-
 +
|-
 +
 +
| [[ File: 1.06.png|thumb |350px|center]]
 +
<br>
 +
<center><b> Features/Metrics </b></br>
 +
 +
||
 +
The metrics of the campaign would include:
 +
 +
1. Goal: What financial target goal should the startup set for the crowd-funding campaign?
 +
 +
2. Number of Award Levels: For each project, the creator of the campaign would be able to set the reward level (e.g. $25, $100, $1000…) for the backers. The question we would want to answer is: How many award levels should the founders set for their project?
 +
 +
3. Average Reward Amount: Similar to metric (2), the founders would be able to set the value of the reward. The question here that we would want to address is: What should the reward amount be?
 +
 +
4. Number of Updates: For every project on Kickstarter, the founders would be giving constant updates to the backers about the progress of the project. We would want to see: How many updates are most common for the campaigns on Kickstarter?
 +
 +
5. Title Length: Other than the image of the campaign, the first impression of the user to the campaign would be the title. Having a concise while at the same time meaningful has proven to be a challenge for many campaign creators. Hence, we would want to see: What is the most common length of the project title?
 +
 +
6. Campaign’s Duration: Last but not least, one crucial determinant of whether the project is successful or not is actually the duration of the campaign. Longer duration gives the investors more time to review the project. However, too long duration would discourage investors as there is no sense of urgency. Hence, the question that we would want to address is: What is the best duration for the campaign?
 +
 +
 +
|-
 +
 +
| [[ File:1.07.png|thumb|350px|center]]
 +
<br>
 +
<center><b> Bean plot: Category (Goal amount vs Success/Failure)  </b></br>
 +
 +
||
 +
In-order to visualize the successful and unsuccessful campaigns vs the goal amount set among the startups, we use a Bean plot across all the categories. Irrespective of all categories, when the goal amount set is from $1000 - $5000 then, the startups have high chances of being successful. Therefore, for small projects this is a great insight to notice and invest in their respective categories. 
 +
When the goal amount set is higher than $10,000 then the startups are more likely being unsuccessful in raising the funds.
 +
 +
 +
|-
 +
 +
| [[ File:1.08.png|thumb|350px|center]]
 +
<br>
 +
<center><b> Bean plot: Sub Category: Music (Goal amount vs Success/Failure) </b></br>
 +
 +
||
 +
Using a interactive drill down Tree-map and when we observe the interactive bean plot for Category (Music) has the second highest number of campaigns in total. Here, a strong insight has been found. Startups whose goal amount set to $10,000 have very high chances of failing to raise the money. So, we conclude in Music category, it is recommended to set below $10,000 to avoid failure.
 +
 +
|-
 +
 +
|-
 +
 +
| [[File:1.09.jpg|thumb|350px|center]]
 +
<br>
 +
<center><b> Word Cloud </b></br>
 +
 +
||
 +
Future Work: Can utilize word cloud to find interesting insights of this massive word data and try to build a visualization effectively to classify successful and un-successful campaigns.
 +
 +
|-
 +
|}
 +
<br>
  
 
<br>
 
<br>
Line 97: Line 228:
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
! Survey of Related Work
+
! '''SURVEY OF RELATED WORK'''
 
|-
 
|-
 
|}
 
|}
Line 119: Line 250:
 
  || This chart shows another point to consider which is how the number of backers affect the success rate of a project. The use of pie chart however, may not be what the group will use.
 
  || This chart shows another point to consider which is how the number of backers affect the success rate of a project. The use of pie chart however, may not be what the group will use.
 
|-
 
|-
| Example || Example
+
| <insert> || <insert>
 
|}
 
|}
 
<!-- End of Related Work -->
 
<!-- End of Related Work -->
Line 125: Line 256:
 
<br>
 
<br>
  
<!-- Start of Charts -->
+
<!-- Start of Technical Challenges -->
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
! Charts
+
! '''TECHNICAL CHALLENGES'''
 
|-
 
|-
 
|}
 
|}
 +
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
! Chart !! Description and Analysis
+
! Technical Challenge !! Solution
 +
|-
 +
| Group is not that experienced and familiar with visualization softwares ||
 +
* Watch Videos Online
 +
* Learn as a group and help each other
 +
|-
 +
| '''Crawling''' <br> Certain information such as the city or state of the campaign and the description of the campaign have to be obtained by crawling ||
 +
* Learn how to crawl online
 +
* Set aside time every day to work together
 
|-
 
|-
|
+
| No experience with javascript and D3 ||  
<center>'''Visualisation Dashboard'''</center>
+
* Watch videos and read about javascript and D3
 +
* Work and ask one another
 +
|}
  
||
+
{| class="wikitable"
[[File:KickFinisher Visualisation Dashboard.png|800px|frameless|center]]
+
|-
<br>
+
! '''IDEAS BRAINSTORMING '''
'''Methodology'''
 
* We would be using most of the data given from the dataset
 
* How the visualisation dashboard works is that, as the user interact with the tree map on the left hand side, the visualisation on the right hand side would be dynamically changed showing the users insights according to the users' interaction with the tree map
 
'''Purpose'''
 
* This is to allow users to explore the world of KickStarter in a lot more intuitive way
 
'''Pros'''
 
* Clearly show the data with all the necessary stats needed before anyone launch their kickstarter campaign
 
'''Cons'''
 
* Information might be cluttered, experimentation is needed
 
 
|-
 
|-
|
 
<center>'''Word Cloud'''</center>
 
||
 
[[File:Word-cloud.jpg|frameless|center]]
 
'''Data Used'''
 
* Titles of projects
 
* ''And other filtering attributes''
 
'''Methodology'''
 
* We would be doing text analysis for all the titles of the projects in each category
 
* Based on the analysis, we would be visualising the result using Word Cloud
 
* There will be filtering function to allow users to only select successful/failed/cancelled projects as well as filtering based on countries/project categories
 
'''Purpose'''
 
* This is to allow the user to understand for each of the category, what are the words being used by the past projects
 
'''Pros'''
 
* It reveals the distribution of words in a visually appealing way
 
* It is engaging with the users
 
'''Cons'''
 
* Users might have difficulty in deciphering the size and comparing the size of one word with another
 
 
|}
 
|}
<!-- End of Charts -->
+
<div style="font-family:Helvetica;font-size:12px">
 +
 
 +
{| class="wikitable" width="100%"
  
<br>
+
|-
 +
| [[File:1.10.jpg|thumb|350px|center]]
 +
|-
  
<!-- Start of Technical Challenges -->
 
{| class="wikitable"
 
 
|-
 
|-
! Technical Challenges
+
| [[File:1.11.jpg|thumb|350px|center]]
 +
 
 
|-
 
|-
 
|}
 
|}
Line 183: Line 301:
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
! Technical Challenge !! Solution
+
! '''APPLICATION DESIGN & TECHNOLOGIES USED '''
 
|-
 
|-
| Group is not that experienced and familiar with visualization softwares ||
+
|}
* Watch Videos Online
+
<div style="font-family:Helvetica;font-size:12px">
* Learn as a group and help each other
+
 
 +
{| class="wikitable" width="100%"
 +
 
 
|-
 
|-
| '''Crawling''' <br> Certain information such as the city or state of the campaign and the description of the campaign have to be obtained by crawling ||  
+
| [[File:1.12.png|thumb|600px|center]]
* Learn how to crawl online
 
* Set aside time every day to work together
 
 
|-
 
|-
| No experience with javascript and D3 ||
+
 
* Watch videos and read about javascript and D3
 
* Work and ask one another
 
 
|}
 
|}
  
 
<br>
 
<br>
 
<br>
 
<br>

Latest revision as of 23:47, 25 November 2018


Kick Finisher Logo 2.png

Proposal

PROBLEM & MOTIVATION

Problem: For many, acquiring enough finances to turn their idea into a business is a big issue so some would go for crowdfunding platforms as a means to obtain the required amount to start their business. Kickstarter is one of the more popular crowdfunding platform, but they have an "all or nothing" rule where the campaign cannot get the donated amount if it does not reach the goal set by the business. As of October 5th 2018, only about one in every three campaigns reach their funding goal, which means that most do not get the funding they need from Kickstarter.

Motivation: There is a need to understand what makes a successful campaign. Are there any attributes that sets apart projects that are successful? The team would like to create several visualizations on kickstarter campaign data to analyze trends and determine if certain attributes of a campaign such as "amount of funding set" affects its chances of succeeding. In the end, the team hopes that others can use the visualizations created in order to have a better understanding of how they should create their campaign to have a higher chance of success.


DATA

The group will only use one csv file as the source of data.

Dataset
ks-projects-201801.csv from https://www.kaggle.com/kemical/kickstarter-projects#ks-projects-201801.csv

Details of Dataset

Column Description
ID ID of the campaign / project
name Name(Title) of the campaign / project
category Sub-category of the campaign
main_category Main category of the campaign
currency Currency used for funds of the campaign
deadline Until when the campaign will run
goal Goal amount to reach of the campaign in their currency
launched When this campaign started
pledged Amount pledged by donors in the currency of the campaign
state Status of the campaign (successful? failed?)
backers How many donors
country Which country is the campaigned launched
usd pledged Pledged amount converted to USD by KS conversion
usd_pledged_real Pledged amount converted to USD by fixer.io
usd_goal_real Goal amount converted to USD


VISUALISATION DASHBOARD & CHARTS
Chart Description and Analysis
Visualisation Dashboard
1.01.png


Drill down Tree-map : Main Category

The first visualization that we are going with is the “Drill-down Tree-map” with the size being the number of campaigns and the color being the category. The visualization showcases all the Main-Categories and Sub-Categories with their respective size being the number of campaigns. The main category is stated at the top left corner with the number while the sub-categories are faded in the background. The visualization contains the data of campaigns between 2015 and 2018 – which indicates the recent trends on the Kickstarter platform.

1.02.png


Drill down Tree-map : Sub Category

This example visualization showcases the number of campaigns for each of the sub-categories of the “Film & Video”. From here we can see that under “Film & Video”, “Short Film” and “Documentary” are the most popular. We can interact with the tree-map and have the same exploration with other main categories as well such as “Music”, “Publishing”, “Theater” …

1.03.png


Stacked Bar Chart

In the stacked bar graph, we use the data across all categories from the year 2015 -2018. We calculate the percentage of successful projects and failed projects for each of the state. The states are then sorted based on the percentage of success in a descending order. From the visualization above we can see that the top 5 states with the highest percentage of successful projects are “New York”, “Rhode Island”, “Vermont”, “Massachusetts” and “Montana”.

1.04.png
1.05.png


Interactive Visualization

The interactivity of the visualization comes in when we interact with the tree-map. After every selection of main- category or sub-category of the tree-map, the bar-graph visualization would be filtered based on the selection.

From the visualization above we can see that even though across all categories, “New York” has the highest percentage of successful projects, however, if we zoom into “Publishing” category, “West Virginia” is a better place to launch the campaign as compared to “California” as it has a higher percentage of successful projects specifically for the “Publishing” category. The above is just one example of the interactivity of the visualization. The user would be able to do the same thing and view the top 5 states with the highest percentage of success for any main-category or sub-category that they select on the tree-map.

1.14.png



From this visualization, we can see that “Dance”, “Theater” and “Music” have the highest percentage of success if we look at the overall data across all the states. However, if the user selects “Minnesota” as the state that they are interested in, they would get the following:

1.15.png



From this, we can see that unlike other states, Minnesota is more popular with “Music” and “Comics”. From the user perspective who are thinking about their project idea to launch in Minnesota, a possible insight from this is that they should take a look at the category of “Music” and “Comics” since these two have relatively high percentage of success.

1.06.png


Features/Metrics

The metrics of the campaign would include:

1. Goal: What financial target goal should the startup set for the crowd-funding campaign?

2. Number of Award Levels: For each project, the creator of the campaign would be able to set the reward level (e.g. $25, $100, $1000…) for the backers. The question we would want to answer is: How many award levels should the founders set for their project?

3. Average Reward Amount: Similar to metric (2), the founders would be able to set the value of the reward. The question here that we would want to address is: What should the reward amount be?

4. Number of Updates: For every project on Kickstarter, the founders would be giving constant updates to the backers about the progress of the project. We would want to see: How many updates are most common for the campaigns on Kickstarter?

5. Title Length: Other than the image of the campaign, the first impression of the user to the campaign would be the title. Having a concise while at the same time meaningful has proven to be a challenge for many campaign creators. Hence, we would want to see: What is the most common length of the project title?

6. Campaign’s Duration: Last but not least, one crucial determinant of whether the project is successful or not is actually the duration of the campaign. Longer duration gives the investors more time to review the project. However, too long duration would discourage investors as there is no sense of urgency. Hence, the question that we would want to address is: What is the best duration for the campaign?


1.07.png


Bean plot: Category (Goal amount vs Success/Failure)

In-order to visualize the successful and unsuccessful campaigns vs the goal amount set among the startups, we use a Bean plot across all the categories. Irrespective of all categories, when the goal amount set is from $1000 - $5000 then, the startups have high chances of being successful. Therefore, for small projects this is a great insight to notice and invest in their respective categories. When the goal amount set is higher than $10,000 then the startups are more likely being unsuccessful in raising the funds.


1.08.png


Bean plot: Sub Category: Music (Goal amount vs Success/Failure)

Using a interactive drill down Tree-map and when we observe the interactive bean plot for Category (Music) has the second highest number of campaigns in total. Here, a strong insight has been found. Startups whose goal amount set to $10,000 have very high chances of failing to raise the money. So, we conclude in Music category, it is recommended to set below $10,000 to avoid failure.

1.09.jpg


Word Cloud

Future Work: Can utilize word cloud to find interesting insights of this massive word data and try to build a visualization effectively to classify successful and un-successful campaigns.



SURVEY OF RELATED WORK
Related Work Points to Consider

The inforgraphic has presented various ideas to explore in terms of what kinds of visualizations to create. The inforgraphic mostly presented numbers and not an in depth visualization.
The things to consider by looking at this work would be the following:

  • By how much do projects/campaigns succeed or fail?
  • What are the success rates per category?
  • Which features of a campaign are related with its success?
This chart shows another point to consider which is how the number of backers affect the success rate of a project. The use of pie chart however, may not be what the group will use.
<insert> <insert>


TECHNICAL CHALLENGES
Technical Challenge Solution
Group is not that experienced and familiar with visualization softwares
  • Watch Videos Online
  • Learn as a group and help each other
Crawling
Certain information such as the city or state of the campaign and the description of the campaign have to be obtained by crawling
  • Learn how to crawl online
  • Set aside time every day to work together
No experience with javascript and D3
  • Watch videos and read about javascript and D3
  • Work and ask one another
IDEAS BRAINSTORMING
1.10.jpg
1.11.jpg
APPLICATION DESIGN & TECHNOLOGIES USED
1.12.png