صياغة أسئلة EDA

# Vague question
"Do student test scores vary by gender?"

# Layer 1 — scope the variable
"What is the distribution of math scores across the dataset?"

# Layer 2 — introduce the comparison
"How do math score distributions differ between male and female students?"

# Layer 3 — add depth
"Does the gender gap in math scores change across different ethnic groups?"

# Layer 4 — actionable insight
"Which subgroups show the largest performance gaps, and what interventions might address them?"

# Example for Students Performance dataset

primary = "What factors most strongly predict student performance on exams?"

secondary = [
    "How do scores distribute across subjects (math, reading, writing)?",
    "Do gender and ethnicity interact to produce score differences?",
    "What role does parental education level play?",
    "Is test preparation course completion associated with higher scores?",
]

tertiary = [
    "Does lunch type (standard vs free/reduced) confound the ethnicity effect?",
    "Are the relationships linear or do they plateau?",
    "Do outliers represent data entry errors or genuine high performers?"
]

# Weak hypothesis (not testable)
"Gender affects scores."

# Strong hypothesis (testable, specific)
"Female students score higher on reading and writing, while male students score higher on math, with effect sizes greater than 0.2 standard deviations."

# Null hypothesis (for confirmatory analysis)
"There is no statistically significant difference in mean math scores between male and female students (p > 0.05)."

eda_questions = {
    "primary": "What factors most strongly predict student performance?",
    "secondary": {
        "score_distributions": "How do scores distribute across subjects?",
        "gender_effects": "Do gender and ethnicity interact?",
        "parental_education": "What role does education level play?",
        "test_prep": "Is test preparation associated with higher scores?",
    },
    "status": {}  # track which questions you've answered
}

for q in eda_questions["secondary"]:
    print(f"  [ ] {q}")

# Your turn: frame EDA questions for a dataset of your choice
# Replace the example below with your own

my_dataset = "Students Performance in Exams"

primary_question = "What factors most strongly predict student performance?"
secondary_questions = [
    "How do scores distribute across subjects?",
    "Do demographic factors correlate with performance gaps?",
    "What is the effect of test preparation course completion?",
]

# Convert one to a hypothesis
hypothesis = (
    "Students who complete a test preparation course will score "
    "at least 5 points higher on average across all three subjects."
)

print(f"Primary: {primary_question}")
print("Secondary:")
for i, q in enumerate(secondary_questions, 1):
    print(f"  {i}. {q}")
print(f"\nHypothesis: {hypothesis}")