Create pipeline /eda.py

src /pipeline /eda.py

@@ -0,0 +1,104 @@
+"""
+eda.py
+------
+Exploratory Data Analysis for the Student Performance Prediction System.
+Generates distribution plots and a correlation heatmap using
+Matplotlib and Seaborn, and prints basic statistical summaries
+(mean, median, std, skew) using NumPy/Pandas.
+
+Run: python3 src/eda.py
+Outputs PNG charts into assets/
+"""
+
+import os
+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+import seaborn as sns
+
+from data_preprocessing import run_pipeline, NUMERIC_FEATURES, TARGET_COLUMN
+
+sns.set_theme(style="whitegrid")
+ASSETS_DIR = "assets"
+
+
+def summarize_statistics(df: pd.DataFrame) -> pd.DataFrame:
+    """Basic probability & statistics summary for numeric features."""
+    cols = NUMERIC_FEATURES + [TARGET_COLUMN]
+    cols = [c for c in cols if c in df.columns]
+    summary = pd.DataFrame({
+        "mean": df[cols].mean(),
+        "median": df[cols].median(),
+        "std_dev": df[cols].std(),
+        "skew": df[cols].skew(),
+        "min": df[cols].min(),
+        "max": df[cols].max(),
+    })
+    return summary.round(2)
+
+
+def plot_target_distribution(df: pd.DataFrame) -> None:
+    plt.figure(figsize=(8, 5))
+    sns.histplot(df[TARGET_COLUMN], kde=True, color="#4C72B0", bins=25)
+    plt.title("Distribution of Final Exam Scores")
+    plt.xlabel("Final Exam Score")
+    plt.ylabel("Number of Students")
+    plt.tight_layout()
+    plt.savefig(os.path.join(ASSETS_DIR, "score_distribution.png"), dpi=150)
+    plt.close()
+
+
+def plot_correlation_heatmap(df: pd.DataFrame) -> None:
+    cols = NUMERIC_FEATURES + [TARGET_COLUMN]
+    cols = [c for c in cols if c in df.columns]
+    corr = df[cols].corr()
+
+    plt.figure(figsize=(9, 7))
+    sns.heatmap(corr, annot=True, fmt=".2f", cmap="coolwarm", square=True, cbar_kws={"shrink": 0.8})
+    plt.title("Feature Correlation Heatmap")
+    plt.tight_layout()
+    plt.savefig(os.path.join(ASSETS_DIR, "correlation_heatmap.png"), dpi=150)
+    plt.close()
+
+
+def plot_study_hours_vs_score(df: pd.DataFrame) -> None:
+    plt.figure(figsize=(8, 5))
+    sns.scatterplot(data=df, x="study_hours_per_day", y=TARGET_COLUMN,
+                     hue="performance_band", palette="viridis", alpha=0.7)
+    plt.title("Study Hours vs Final Exam Score")
+    plt.xlabel("Study Hours per Day")
+    plt.ylabel("Final Exam Score")
+    plt.tight_layout()
+    plt.savefig(os.path.join(ASSETS_DIR, "study_hours_vs_score.png"), dpi=150)
+    plt.close()
+
+
+def plot_performance_band_counts(df: pd.DataFrame) -> None:
+    plt.figure(figsize=(7, 5))
+    order = ["Needs Improvement", "Average", "Good", "Excellent"]
+    sns.countplot(data=df, x="performance_band", order=order, hue="performance_band",
+                  palette="Blues_d", legend=False)
+    plt.title("Student Count by Performance Band")
+    plt.xlabel("Performance Band")
+    plt.ylabel("Count")
+    plt.tight_layout()
+    plt.savefig(os.path.join(ASSETS_DIR, "performance_band_counts.png"), dpi=150)
+    plt.close()
+
+
+if __name__ == "__main__":
+    os.makedirs(ASSETS_DIR, exist_ok=True)
+    df = run_pipeline("data/student_data.csv")
+
+    print("=== Statistical Summary ===")
+    print(summarize_statistics(df))
+
+    print("\n=== Correlation with target (final_exam_score) ===")
+    cols = NUMERIC_FEATURES + [TARGET_COLUMN]
+    print(df[cols].corr()[TARGET_COLUMN].sort_values(ascending=False).round(3))
+
+    plot_target_distribution(df)
+    plot_correlation_heatmap(df)
+    plot_study_hours_vs_score(df)
+    plot_performance_band_counts(df)
+    print(f"\nSaved 4 charts to {ASSETS_DIR}/")