Page 13 - Playtest_Square_Enix - EphemeralData

The Frustration can be defined as : "inability to achieve a goal or desire"

We will try to approximate the feeling of frustration by doing as follow :

Frustration - Fail Frustration - Quit

Total fail Total quit
________________ ________________

Total pass Total pass

Frustration - Fail: Considering all the pass and fail, how many times on
average a user had to fail to pass once. The higher this feature, the
higher is the likelihood for a user to be frustrated because of failure.

Frustration - Quit: Considering all the pass and quit, how many times on
average a user had to quit to pass once. The higher this feature, the
higher is the likelihood for a user to be frustrated because of quitting.

These two features can be considered good indicators of frustration, and
even let us split the potential frustration coming from failing or quits.

Moving Average - Moving Average -

Frustration - Fail Frustration - Quit

Here we are computing a moving average of two periods on the two
features measuring the frustration. By doing so, we will be able to see
how the global frustration shifted through checkpoints (i.e. has the

frustration increased or decreased smoothly or brutally).

We took a moving average of two periods 2 reasons :

- By taking into account the previous period we can integrate a part of
the residual frustration coming from the previous checkpoint into the
next.

- Using 2 periods let us compute the moving average, while only losing
the first checkpoint value. Indeed, we don't have a previous period while

in checkpoint #1 so the moving average can't be computed.

12