EduExcluders - Explainable Artificial Intelligence

Notebooks

The code is seperated into various notebooks, namely:

• eda_before_preprocessing.ipynb contains the exploratory data analysis notebook.
• data_preprocessing.ipynb contains the notebook which preprocesses the data.
• model_testing.ipynb contains the various models that were used during EDA and after to find a good option.
• LIME.ipynb contains the LIME explanations
• SHAP.ipynb contains the SHAP explanations
• PDP_ICE.ipynb contains the PDP/ICE explanations

Dataset

This study uses the Predict Students' Dropout and Academic Success dataset from the UC Irvine Machine Learning Repository. The dataset is located in the data directory of this repository, in which data.csv is the original data and formatted_data is the preprocessed data.

Dataset info

Variable Name	Role	Type	Demographic	Description	Units	Missing Values
Marital Status	Feature	Integer	Marital Status	1 – single 2 – married 3 – widower 4 – divorced 5 – facto union 6 – legally separated		no
Application mode	Feature	Integer		1 - 1st phase - general contingent 2 - Ordinance No. 612/93 ...		no
Application order	Feature	Integer		Application order (between 0 - first choice; and 9 last choice)		no
Course	Feature	Integer		33 - Biofuel Production Technologies 171 - Animation and Multimedia Design ...		no
Daytime/evening attendance	Feature	Integer		1 – daytime 0 - evening		no
Previous qualification	Feature	Integer	Education Level	1 - Secondary education 2 - Higher education - bachelor's degree ...		no
Previous qualification (grade)	Feature	Continuous		Grade of previous qualification (between 0 and 200)		no
Nationality	Feature	Integer	Nationality	1 - Portuguese; 2 - German; 6 - Spanish; ...		no
Mother's qualification	Feature	Integer	Education Level	1 - Secondary Education - 12th Year of Schooling or Eq. ...		no
Father's qualification	Feature	Integer	Education Level	1 - Secondary Education - 12th Year of Schooling or Eq. ...		no
Mother's occupation	Feature	Integer	Occupation	0 - Student 1 - Representatives of the Legislative Power and Executive Bodies, Directors, Directors and Executive Managers ...		no
Father's occupation	Feature	Integer	Occupation	0 - Student 1 - Representatives of the Legislative Power and Executive Bodies, Directors, Directors and Executive Managers ...		no
Admission grade	Feature	Continuous		Admission grade (between 0 and 200)		no
Displaced	Feature	Integer		1 – yes 0 – no		no
Educational special needs	Feature	Integer		1 – yes 0 – no		no
Debtor	Feature	Integer		1 – yes 0 – no		no
Tuition fees up to date	Feature	Integer		1 – yes 0 – no		no
Gender	Feature	Integer	Gender	1 – male 0 – female		no
Scholarship holder	Feature	Integer		1 – yes 0 – no		no
Age at enrollment	Feature	Integer	Age	Age of student at enrollment		no
International	Feature	Integer		1 – yes 0 – no		no
Curricular units 1st sem (credited)	Feature	Integer		Number of curricular units credited in the 1st semester		no
Curricular units 1st sem (enrolled)	Feature	Integer		Number of curricular units enrolled in the 1st semester		no
Curricular units 1st sem (evaluations)	Feature	Integer		Number of evaluations to curricular units in the 1st semester		no
Curricular units 1st sem (approved)	Feature	Integer		Number of curricular units approved in the 1st semester		no
Curricular units 1st sem (grade)	Feature	Integer		Grade average in the 1st semester (between 0 and 20)		no
Curricular units 1st sem (without evaluations)	Feature	Integer		Number of curricular units without evaluations in the 1st semester		no
Curricular units 2nd sem (credited)	Feature	Integer		Number of curricular units credited in the 2nd semester		no
Curricular units 2nd sem (enrolled)	Feature	Integer		Number of curricular units enrolled in the 2nd semester		no
Curricular units 2nd sem (evaluations)	Feature	Integer		Number of evaluations to curricular units in the 2nd semester		no
Curricular units 2nd sem (approved)	Feature	Integer		Number of curricular units approved in the 2nd semester		no
Curricular units 2nd sem (grade)	Feature	Integer		Grade average in the 2nd semester (between 0 and 20)		no
Curricular units 2nd sem (without evaluations)	Feature	Integer		Number of curricular units without evaluations in the 2nd semester		no
Unemployment rate	Feature	Continuous		Unemployment rate (%)		no
Inflation rate	Feature	Continuous		Inflation rate (%)		no
GDP	Feature	Continuous		GDP		no
Target	Target	Categorical		Target. The problem is formulated as a three category classification task (dropout, enrolled, and graduate) at the end of the normal duration of the course		no