Page 19 - Playtest_Square_Enix - EphemeralData

Checkpoints' Segmentation

Objectives :

T he segmentation of the checkpoint is not the most useful analysis

when used alone. However when used as a support analysis, it can

help and reinforce results found via other methods.

In our case we want to see if some checkpoints share similar traits so
that if we use statistic algorithms without human interaction, will we
have interesting segmented groups.

Clustering Methodology:

All features available

reduce our 30 features
Principal Component Analysis into only 2 ones

Using our 2 aggregated
K-Means Algorithm features, will cluster

checkpoints in 3 groups

After these processes, we obtained a segmentation totally done by
statistics with no human meddling.

What is already striking before any analysis, is how the checkpoint #3 is
an outlier on this segmentation. Even though it is from the blue group, it
is far away from the checkpoints of its group.

Index of Dedication
CP#3

CP#15
CP#5
CP#17 Replayability
CP#2 Ind. Frustration
CP#21 CP#6 CP#1
CP#25 CP#7
CP#19 CP#13
CP#27 CP#23 CP#9 CP#4
CP#11
CP#16 CP#14 CP#8
CP#29 CP#20 CP#10
CP#22
CP#12
CP#26 CP#24 CP#18
CP#28 Perseverance Tenacity
18