Skip to content

W2201 Data Model For Reporting

Jump to bottom
Richard Burgmann edited this page Jun 14, 2017 · 2 revisions

Reporting

Introduction

What I want to learn determines what data I need to learn it.

Things to learn / questions to answer.

A Benchmark Measure

I need a benchmark to compare AI strategies to, for this I will use a random walk through the cave system. My hypothesis is that the further the Adventurer is from the Gold the lower their overall score as they are more likely to encounter a hazard. The secondary hypothesis is that the reduction in the score compared to distance will drop by a factor of 1/4. As the Adventurer can only move in the cardinal directions the distance measure will be the manhattan distance.

I think two files need to be created.
The first contains the parameters in place at the time of the experiment (call this the parameter file).

  • Date of the experiment.
  • The parameters for placing the entities within the grid (Fixed or Random)
  • If fixed then the values of the locations.
  • The starting health of the Adventurer.

The second contains the results of the experiment (call this the data file).
The data file should have for each run the following information;

  • The starting location of the Adventurer.
  • The location of the gold.
  • A boolean flag indicating if the Adventurer died from falling into a pit.
  • A boolean flag indicating if the Adventurer died from being eaten by the Wumpus.
  • A boolean flag indicating if the Adventurer died from starvation.
  • A boolean flag indicating if the Adventurer found the gold.
  • The final score for the Adventurer.
Clone this wiki locally