Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.


MSiA 2013 Cohort

Data Visualization Final Project

Shawna Baskin

on 24 July 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of MSiA 2013 Cohort

Over represented
Under represented
For both assigned and chosen groups, we plotted the density distribution for percent of group members with a specified characteristic to approximate the underlying probability density function for the specified characteristic and group type combination.

We identified stronger tails for chosen groups than for assigned groups, for the majority of the characteristics. This indicates people tend to prefer working with other people who are similar to themselves. The one characteristic inconsistent with this trend was the S/N Myers-Briggs characteristic, which we also noticed in our analysis on group member similarity. For this trait, we noted a much higher percentage of chosen groups with close to an even distribution between "Sensing" and "Intuition" members than for assigned groups.
Analyzing the MSiA 2013 Cohort
Teammate Selection Networks
To help us look for patterns in teammate selection, we chose to visualize repeat partnerships as a social network. Each node is a member of the cohort, and a link represents if worked together. Nodes were colored based on a variety of attributes, in this example we used the Myers-Briggs judging/perceiving dimension. Our network is very dense when we look at any repeat partnerships, but patterns become more evident when we look at pairs who worked together for three or more projects.

We also visualized the shared attributes between specific pairs of collaborators using a chord diagram. Similar to the network nodes were colored by attribute. In this one, links are colored by the attribute if both teammates are the same and grey if the two people have different qualities.
Teammate Selection Networks
Two or More Group Projects Together
Teammate Selection Networks
Three or More Group Projects Together
What Groups do
We Choose?
We wanted to better understand the criteria used when groups select their members, so we compared the composition of chosen groups with the composition of assigned groups for the MSIA 2012/2013 cohort. We calculated group composition variables based on gender, Myers-Briggs personality type (E/I, S/N, T/F, J/P), degree category (math and engineering/non-math and engineering), work experience (one or more years/none), location (Evanston/Chicago), Origin (US/International), and self-rated area of highest skills (technical/soft skills)
Group Composition
Group Member Similarity
Parallel Coordinates Plot
We used a parallel coordinates plot to compare the percent of group members who are the same on each characteristic for chosen groups versus our baseline, assigned groups. We broke down chosen groups into team sizes of only 3 or 4 members (to more similarly reflect the composition of the assigned groups) as well as all team sizes, which includes team sizes of 2 through 7 people.

We determined groups choose members that are similar at a higher rate than assigned groups in all areas except the S/N Myers-Briggs characteristic. We believe having both "Sensing" and "Intuition" team members to be the most important of all examined characteristics for a successful data science team.
Skills Distribution
The distribution of responses varied based on the type of skill - data, programming and soft.
How Do We Perceive Our Skills?
The histograms show the skill level that students rated themselves relative to the entire MSIA cohort.
Soft Skills
Programming Skills
Students tend to be under-confident in their programming abilities.
Data Skills
Students tend to be slightly over-confident in their data tool skills.
Students tend to be over-confident in their soft skills.
Conducted a survey of all 30 Northwestern Masters in Analytics students with questions on:
Shawna Baskin, Katharine Matsumoto,
Laura Siahaan, Alice Zhao
Myers-Briggs Type
Project Teams
Skills Evaluation

Website Link:
Group Composition
Density Plots
What are the Myers Briggs
Personality Types?
Source: Washington Post
93% of the MSiA class feel that the Myers Briggs Types are an accurate representation of themselves.
How do we stack up?
The MSiA class is very skewed in its distribution of Sensing/Intuition and Thinking/Feeling. Compared to the population, the cohort leans toward Intuition and Thinking. These traits are the foundational elements of core personality components.
People prefer working with people
who are similar to themselves
Self-selected groups have less diversity
Group Member Similarity
Who Works with Who?
All Group Projects
See More: http://www.matsumoto.com/KMM/analytics/Ladies1.html
See More: http://www.matsumoto.com/KMM/analytics/Ladies1.html
Who are we really?
Reverse Engineering
Personality Traits
A principle component analysis reveals a chart similar to the previous square. The first two principle components explain 67% of one's Myers Briggs personality. Intuition/Sensing are a strong factor in both components.
Checkout the website for an interactive 3d cube!
The Myers Briggs Personality Types
The Myers Briggs personality types are popular tool for organizations to better understand their members and improve workplace relationships. Compared to the general population, the MSiA cohort has an over representation of Thinking and Intuition types. A common way to show the the Myers Briggs types is in a square. We wanted to see how our square compares. The MSiA cohort gravitates towards Possible and Logical along the 45° line connecting ST and NF.

To better understand the reasoning behind the positioning of personality types within the square, we conducted a Principle Components Analysis which revealed underlying correlations that define each types spacial positioning. In the PCA Analysis, S/N and T/F represent the top loading factors for PC1. This is in line with the standard square. Overall, the PCA analysis provided a similar output to the square with a few key differences. Many students in our class lie on the cusp of two different types such as N vs. S. This scalar view allows us to see moderates vs. extremes. There are more students that gravitate towards the center of each type.
Full transcript