ANALYSIS PLAYGROUND

1

0

2

3

Excluding the data from 5 cm away or closer to the checkpoints do not change the significance of the factors.

Factor (0) F-Value p-value
Group 4.88 0.035
OF 1.97 0.171
Interaction 3.69 0.065
Factor (1) F-Value p-value
Group 6.31 0.018
OF 21.06 < 0.000
Interaction 4.66 0.040
Factor (2) F-Value p-value
Group 9.91 0.003
OF 0.79 0.383
Interaction 0.18 0.671
Factor (3) F-Value p-value
Group 10.45 0.003
OF 0.311 0.582
Interaction 1.48 0.23

1

0

2

3

Factor (0) F-Value p-value
Group 4.88 0.130
OF 1.97 0.431
Interaction 3.69 0.190
Factor (1) F-Value p-value
Group 6.31 0.050
OF 21.06 < 0.000
Interaction 4.66 0.10
Factor (2) F-Value p-value
Group 9.91 0.026
OF 0.79 0.373
Interaction 0.18 0.729
Factor (3) F-Value p-value
Group 10.45 0.019
OF 0.311 0.546
Interaction 1.48 0.617

Excluding

BD01

LC01

BH02

Factor (full) F-Value p-value
Group 10.54 0.003
OF 6.77 0.015
Interaction 5.15 0.031
Factor (travel) F-Value p-value
Group 11.99 0.002
OF 5.25 0.030
Interaction 2.23 0.14
Factor (trial) F-Value p-value
Group 11.56 0.002
OF 0.58 0.452
Interaction 0.84 0.369
Factor (full) F-Value p-value
Group 10.54 0.021
OF 6.77 0.036
Interaction 5.15 0.088
Factor (travel) F-Value p-value
Group 11.99 0.016
OF 5.25 0.020
Interaction 2.23 0.227
Factor (trial) F-Value p-value
Group 11.56 0.021
OF 0.58 0.808
Interaction 0.84 0.639

Excluding

BD01

LC01

BH02

RESULTS

w/o OF

OF

Factor p
Group < 0.01
Optic Flow 0.03
Group x Optic Flow 0.14

w/o OF

OF

  • Control Group is faster
  • Groups are slow in the environment with optic flow
Factor p
Group 0.02
Optic Flow < 0.01
Group x Optic Flow 0.58

RESULTS

w/o OF

OF

w/o OF

OF

  • Control Group is faster
  • Groups are slow in the environment with optic flow
Factor p
Group 0.09
Optic Flow < 0.01
Group x Optic Flow 0.10

RESULTS

w/o OF

OF

w/o OF

OF

  • No difference between groups
  • Accuracy improves in the environment with optic flow

DATASET

Time Series Data

  • Position of the Participant
  • Heading Direction
  • Time
  • Group
  • Optic Flow

DATASET

Glaucoma

Control

DATASET

Model Concerns

  • Time to complete one trajectory is different across participants.
  • Trajectory is similar but the way participants walk is different.
  • Different trajectories
  • Each trajectory consists of 4 trials for each participant

 

 

  • Classification of participants using navigation data

PRELIMINARY RESULTS

200 data points sampled from each trials

1392 trials in total (29 participants * 48 trials)

Model : ResNet 18

Predictors : Location X, Location Y and Heading Direction

Analysis Playground

By Safa Andac

Analysis Playground

  • 57