Making of: Taste, film makers and review sites

14, Dec, 2021

Week-01: Sources + Sketches:

Data of choice:

Work with 3 sources (letterboxd, imdb, metacrit) of movie ratings to see different "tastes" in different genres of movies. Movies are inherently a subjective medium, but we still look for certain reviewers who/where they fit our tastes the most for recommendations.

Identifying potential variables:

Qualitative/Textual

Movie title, short description

RunTime: Quantity:

RunTime, Ratings, YearReleased, NumVotes, GrossRevenue

Variable of interests:

Ratings + Genres + Director + GrossRevenue

Cognitive processing:

Rating across 3 sites: Grouping with directors' body of works.

Audiences:

Started out with a personal interest in different film rating sites' tastes on certain groups of movies, extending it for moviegoers looking for more customized movie ratings, looking to compare metrics across different sites.

Rough sketches:

Week-02 Proposal:

Iteration prototype

Subject ︎︎︎ Movies:

Individual instances
Cautious of unwarranted links(Between instances)
Connections should be made within the same movies across different sites

Main components:

Potential plots:

Rating Sites:

Metacritic - Letterboxd - IMDB:

Categorical variable

Shape
Color
Texture

Box plots
- Individual movies as single bars with 3 nested numbers(3 rating sites)
- Pros- Works well on a high level
- Cons- The nested values doesn't map to value bounds and median

Variables:

Ratings:

-Stand-alone variable. (Top priority)

Size
- Circle Diameter
- Bar height
- Area
- Line thickness
- Pie angle
- 3D-Circle angle crease shades
- Icicle Diagram
Location:
- Axis plots

Genres/Directors:

-Categorical Groupings

Shapes
Textures
Colors

Week-03 Prototype + Feedback:

Feedbacks from the last iteration:

The individual movie instance serves as a good way to show the different ratings across platforms, yet limited the ability to compare with other movies.
The visualization demanded a considerable amount of screen estate to show just one instance, making the small multiples tricky to stack without losing resolution.

Revised iteration:

Inspired by the visualization of "Colours in Culture" below, moving the scale one layer above, extracting one single movie instance into a slice of a circle, storing 3 ratings in each slice on a sliding scale.

Left:

Before prototyping, I was already having skepticism about the overlapping of ratings, and it would turn out that my worries were warranted, leading me to extract the ratings into a flat setting.

Right:

The extracted rating version will prove to be confusing as well, as each movie were all consisted of three identical colors and with relatively similar height and the exact same angular width, making testing audiences hard to differentiate between movies. Yet I would keep the idea of grouping data with directors moving forward.

Week-04 Usability testing & prototype:

From one of my conversations with my flatmate led me to radar charts, which I had considered in a very early stage, later went back to it identifying that it is simply the best option I can be working with right now given the time frame.

The visualization came from the great step by step tutorial from Nadieh Bremer's blog︎︎︎

Thanks to the feedbacks from Yann Kerblat and Ashley, I was able to identify and implement additional potentials for detailed/layered information from interactions. The years in the image above were added after my meeting with Yann, which expressed confusion on the sequences I presented the films.

I later also added the ability to hover and show each platform's average ratings across the filmmaker's whole body of works, which also came from the suggestions from both Yann and Ashley.

Final + Post Mortem:

Presentation/Crit feedbacks:

Zach's feedback helped me identify a giant blind spot for my final iteration: The choice to present the movies in a circle. In which I did realize (and have shown in the previous section) when I was having 1 on 1 discussion with Ashley, we decided that the implication of a whole work with the idea of "body of work". Yet Zach's question on displaying films in a circle came from an intuitive and valid place: The desire to see the "trajectory" or "growth" of that creator. And with my bad English, I almost automatically associate the word "trajectory" with a connection to scatter plot.

This realization of such a prominent blind spot led me to the story telling of data viz, given any scope of variable presented, there will almost always be unlimited ways to frame the narratives, while my intentions was to compare ratings across three platforms, the audiences intuitive interpretation seems to be a more introspective assessment for one individual's quality of work, which served as a reminder to me to take extra steps to frame my narratives.

As for my next step, I'd love to dig deeper in to case studies this visualization had presented, how different platforms have a certain "bias" towards certain filmmakers/age/sex/genres, and the intuitive/self-contained closed loop shape can serve as a great tool identify certain patterns.

ChooseImage