mv22003 / PredictingEmpathy

Predicting empathy score for employee recruitment using eye-tracker technology

In this repository you will find the following:

1. test_rawdata ------> Zip file with the raw data of 30 participants, performing gaze typing.
2. control_rawdata --> Zip file with the raw data of 30 participants, performing free viewing.
3. questionnaries ----> Zip file with two questionnaries before and after the task performed by the participants. This data contains the empathy scores we are predicting

This three zip files were collected from the EyeT4Empathy dataset, and can be found here.

4. feature_extraction.py --> Python file that contains a series of functions that automatizes the feature extraction of both the control and test group of the dataset. This functions are called in a jupyter notebook to create a matrix for each and store them in a csv file. The functions are the following:

• select_group( )--> Automatize choosing between test or control group when doing feature extraction.
• preprocess( )----> Call the functions that perform the feature extraction of one recording.
• label( )-----------> Extract labels/target from questionnaries to append to dataframe.
• flatten( )---------> Aditional function to flatten a list of list into a list (helpful in the creation of label).

5. notebook_feature_extraction.ipynb --> Jupyter notebook where we can extract the features of both the test and control groups, verifying the dataframe and saving them to a csv.

6. models_control.ipynb --> Jupyter Notebook were we call our matrix and perform several regression models to the control group data and compare them to a dummy regressor to understand the performance of each. There are more explanations about the process inside the .ipynb file.

7. models_test.ipynb -----> Jupyter Notebook were we call our matrix and perform several regression models to the test group data and compare them to a dummy regressor to understand the performance of each. There are more explanations about the process inside the .ipynb file.

8. rfe_models_control.ipynb --> Jupyter Notebook that contains the Recursive Feature Elimination (RFE) process using the best predictor from previous classificators for control group.

9. rfe_models_test.ipynb ------> Jupyter Notebook that contains the Recursive Feature Elimination (RFE) process using the best predictor from previous classificators for test group.