FINAL - Githubissues

eockfen / EyeTism

Early detection of Autism Spectrum Disorder (ASD) is crucial for children's development, yet the diagnostic procedure remains challenging. EyeTism employs machine learning on eye tracking data from both high-functioning ASD and typically developing children (TD) to create a diagnostic tool based on their distinct visual attention patterns.

Other

5 stars 1 forks source link

FINAL #66

Closed azabicki closed 6 months ago

azabicki commented 6 months ago

cleaned up the latest results stored final models in "models/FINAL" folder

new notebooks:

thresholding_RF_XGB_SVC_Voting.ipynb
- calibrates basemodes /RF, XGB, SVC) and saves calibrated models
- also creates/saves voring model
- contains thresholding results for the 4 mentioned models
thresholding_stacking.ipynb
- thresholding procedure on calibrated stacking models
- same procedure as before
FINAL.ipynb
- summarizing the results from previous thresholding procedures
- plot performance of models on each image