Transcription
Key Attributes
- honest , it is the worst timing
- quite good today
- perfect for Halloween does color
- favorite shade
- perfect kind
- color goes on extra extra smooth
- terrible application
- looks perfect
- clip who I 've never actually tried this on before so far , so good
- color is really fun
- lips are literally starting
- creamy colors
- little bit yellow , but not too bad nice
- color is so beautiful
- beautiful color
- kind of color , it 's very unique
- little bit more of a very kind of undertone they 're , so beautiful
- kind of undertone they 're , so beautiful , I even love
- perfect , like full makeup
- perfect every day
- really beautiful ! So pretty with this eye makeup
- slightly peachy pink
- incredible on dark skin tones
- neutral color , I 'm terrible
- terrible application
- most beautiful medium brown shade
- dull parts
- how dull parts
- dull pops
- really beautiful kind
- really unique color
- perfect on some other super porcelain skin
- color is so fun
- super vibrant , just the perfect kind
- one is really beautiful
- faulty or if this is actually what the color
- faulty one
- look amazing
- People are always so scared
- favorite colors
- so flattering and one of the colors
- very subtle shimmer
- best application
- nice crisp line of the actual color itself is so stunning
- perfect Christmas color
- color super festive
- application is terrible , but can we appreciate how amazing
- favorite for sure that Unicorn flash that other dark red
- favorite colors
- sexy color
- beautiful shade
- dull green
- easy enough to pair up with a really warm smoky eye
- colorful green
- guys enjoyed it hope it was helpful
- favorite color
No Data Found for Positive breakout for attribute availability
METHODOLOGY
The study is based on research and interpretation through our flagship platform V.O.I.S.E to analyse opinions, facial expressions, vocal tonality and sentiments.
DATA SOURCE
Videos available on online video sharing platforms.
HOW IT WORKS
Videos are selected based on certain keywords resonating with the category.
All inputs are housed in a custom, closed study dashboard (where they are reviewed for quality assurance - videos of low integrity are rejected)
Once all inputs are received, the responses are indexed and attributed via textual, visual and emotional analysis
Analysis features scoring in the form of TFS (True Feel Score) to normalise measurement of opinion and sentiment strength for the attributes. Inputs and TFS are then used to inform initial key implications and insights.
OPERATIONAL DEFINITION OF HIGH ENGAGEMENT AND LOW ENGAGEMENT VIDEOS
Meta-data from the videos are fed into K-Means clustering algorithm. Data Points used for clustering are “likes per view”, “views per subscriber”, “likes per dislike”.
The K means clustering algorithm clusters the videos. Videos with higher Likes per Dislike, Views per Subscriber and Likes per View is clustered into High Engagement cluster whereas videos with Low Likes per Dislike, Views per Subscriber and Likes per View is clustered into Low engagement videos.
TFS (TRUE FEEL SCORE) PROXIMITY METHODOLOGY
This is measured on a scale of 0-10 where closest proximity is '10' and farthest is '0'.
Opinion polarity has two dimensions: Positive and Negative.
We have scored words that convey an array of feelings. The words are scored based on the context of conversation and delivery.