Update app.py

app.py

@@ -7,108 +7,122 @@ from sklearn.decomposition import PCA
 from sklearn.preprocessing import StandardScaler
 from docx import Document
 from docx.shared import Inches
+from scipy.stats import skew
+
+def analyze_nature(df):
+    num_cols = df.select_dtypes(include=[np.number]).columns
+    insights = []
+    
+    for col in num_cols:
+        # 1. Symmetry/Skewness check
+        s = skew(df[col].dropna())
+        if abs(s) < 0.5:
+            shape = "Symmetric (Approx. Normal)"
+        elif s > 0.5:
+            shape = "Right-Skewed (Positive)"
+        else:
+            shape = "Left-Skewed (Negative)"
+            
+        # 2. Outlier check (IQR Method)
+        Q1 = df[col].quantile(0.25)
+        Q3 = df[col].quantile(0.75)
+        IQR = Q3 - Q1
+        outliers = df[(df[col] < (Q1 - 1.5 * IQR)) | (df[col] > (Q3 + 1.5 * IQR))]
+        
+        insights.append(f"Attribute '{col}': {shape} | Outliers detected: {len(outliers)}")
+    
+    return "\n".join(insights)
 
 def perform_mining(file):
-    if file is None: 
-        return [None] * 6
+    if file is None: return [None]*6
     
     df = pd.read_csv(file.name)
     num_cols = df.select_dtypes(include=[np.number]).columns.tolist()
     
-    # 1. Dataset Description Summary
-    rows, cols = df.shape
-    mem_usage = df.memory_usage(deep=True).sum() / 1024
-    desc_text = f"Dataset contains {rows} observations and {cols} attributes. Total memory footprint: {mem_usage:.3f} KB."
+    # Dataset Description
+    desc_text = f"Observations: {df.shape[0]} | Attributes: {df.shape[1]}"
     
-    # 2. Statistical Summary - Strict 3 Decimal Correction
-    stats = df.describe().round(3).reset_index()
-    stats.rename(columns={'index': 'Statistic'}, inplace=True)
+    # Data Nature & Statistical Inference
+    nature_report = analyze_nature(df)
     
-    # 3. PCA & Nature of Data
-    nature_insight = ""
-    pca_text = "N/A"
-    if len(num_cols) >= 2:
-        scaler = StandardScaler()
-        df_imputed = df[num_cols].fillna(df[num_cols].mean())
-        df_norm = scaler.fit_transform(df_imputed)
-        pca = PCA(n_components=2)
-        pca.fit(df_norm)
-        var_ratio = pca.explained_variance_ratio_
-        pca_text = f"PC1: {var_ratio[0]:.3f} | PC2: {var_ratio[1]:.3f}"
-        
-        if var_ratio[0] > 0.5:
-            nature_insight = "The data exhibits high multicollinearity; a few principal components capture most of the variance."
-        else:
-            nature_insight = "The data is complex and spherical; variance is spread across multiple dimensions."
+    # Correlation Inference
+    corr_matrix = df[num_cols].corr()
+    high_corr = corr_matrix.unstack().sort_values(ascending=False)
+    high_corr = high_corr[(high_corr < 1) & (high_corr > 0.7)].drop_duplicates()
+    corr_inference = f"Strong Correlations found: {len(high_corr)} pairs."
 
-    # 4. Generating 15-Plot Dashboard
+    # Stats rounded to 3 decimals
+    stats = df.describe().round(3).reset_index()
+    
+    # Plotting 15-Plot Suite
     fig = plt.figure(figsize=(15, 20))
     for i in range(15):
         plt.subplot(5, 3, i+1)
         col = num_cols[i % len(num_cols)]
         if i < 5:
-            sns.histplot(df[col].dropna(), kde=True, color='royalblue')
-            plt.title(f"Histogram: {col}")
+            sns.histplot(df[col].dropna(), kde=True, color='blue')
+            plt.title(f"Dist: {col}")
         elif i < 10:
             sns.boxplot(y=df[col].dropna(), color='cyan')
-            plt.title(f"Boxplot: {col}")
+            plt.title(f"Outliers: {col}")
+        elif i == 10:
+            sns.heatmap(corr_matrix, annot=True, cmap='coolwarm', fmt=".2f")
+            plt.title("Correlogram")
         else:
-            sns.scatterplot(x=df[num_cols[0]], y=df[col], alpha=0.6)
+            sns.scatterplot(x=df[num_cols[0]], y=df[col], alpha=0.5)
             plt.title(f"Scatter: {col}")
 
     plt.tight_layout()
-    plot_path = "mining_dashboard.png"
+    plot_path = "mining_results.png"
     plt.savefig(plot_path)
     plt.close()
 
-    # 5. Professional DOCX Report
+    # Doc Generation
     doc_path = "Sastra_Data_Mining_Report.docx"
     doc = Document()
-    doc.add_heading('Amrita Manthan: Preprocessing Report', 0)
-    doc.add_paragraph(f"Faculty: Prof. B. Santhi | SRC, SASTRA")
-    doc.add_heading('Dataset Description', level=1)
-    doc.add_paragraph(desc_text)
-    doc.add_heading('Inference & Nature of Data', level=1)
-    doc.add_paragraph(nature_insight)
-    doc.add_picture(plot_path, width=Inches(6))
+    doc.add_heading('Amrita Manthan: Data Mining Report', 0)
+    doc.add_paragraph(nature_report)
     doc.save(doc_path)
 
-    return stats, desc_text, plot_path, nature_insight, pca_text, doc_path
+    return stats, desc_text, plot_path, nature_report, corr_inference, doc_path
 
-# --- UI Layout ---
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
-    # Centered Header: Logo then Title
-    with gr.Column(variant="panel"):
-        with gr.Row():
-            gr.Image("logo.jpg", width=120, show_label=False, container=False, scale=1)
-        gr.Markdown("<h1 style='text-align: center; color: #0000FF;'><b>Amrita Manthan</b></h1>")
-        gr.Markdown("<h2 style='text-align: center; color: #0000FF;'><b>Preprocessing & Data Mining Portal</b></h2>")
-        gr.Markdown("<p style='text-align: center; color: #0000FF;'><b>Prof. B. Santhi, SRC, SASTRA</b></p>")
+    # Header: Logo Center, Text Left
+    with gr.Row():
+        with gr.Column(scale=1):
+            pass # Spacer
+        with gr.Column(scale=2):
+            gr.Image("logo.jpg", width=120, show_label=False, container=False)
+        with gr.Column(scale=1):
+            pass # Spacer
+            
+    with gr.Row():
+        with gr.Column():
+            gr.Markdown("<h1 style='color: #0000FF; text-align: left;'><b>Amrita Manthan</b></h1>")
+            gr.Markdown("<h2 style='color: #0000FF; text-align: left;'><b>Preprocessing & Data Mining Portal</b></h2>")
+            gr.Markdown("<h3 style='color: #0000FF; text-align: left;'><b>Prof. B. Santhi, SRC, SASTRA</b></h3>")
 
     with gr.Tabs():
-        with gr.TabItem("1. Input & Theory"):
-            gr.Markdown("### **Preprocessing Formalisms**")
-            gr.Markdown(r"**Standardization Formula:** $$z = \frac{x - \mu}{\sigma}$$")
-            gr.Markdown(r"**PCA Projection:** $$\mathbf{T} = \mathbf{X} \mathbf{W}$$")
-            input_file = gr.File(label="Upload Student Dataset (CSV)")
-            run_btn = gr.Button("EXECUTE DATA MINING", variant="primary")
+        with gr.TabItem("1. Theory & Upload"):
+            gr.Markdown("### **Mathematical Clarity**")
+            gr.Markdown(r"**Skewness ($\gamma_1$):** $$\gamma_1 = E\left[\left(\frac{X-\mu}{\sigma}\right)^3\right]$$")
+            gr.Markdown(r"**Outlier Detection (IQR):** $$[Q1 - 1.5 \times IQR, Q3 + 1.5 \times IQR]$$")
+            u_file = gr.File(label="Upload CSV")
+            run_btn = gr.Button("ANALYZE", variant="primary")
             
-        with gr.TabItem("2. Statistics & Description"):
-            out_desc = gr.Textbox(label="Dataset Description", placeholder="Summary of rows, columns, and memory...")
-            out_stats = gr.DataFrame(label="Descriptive Statistics (3 Decimal Correction)")
+        with gr.TabItem("2. Statistics & Skewness"):
+            gr.Markdown("#### **Dataset Summary**")
+            out_desc = gr.Markdown()
+            out_stats = gr.DataFrame(label="3-Decimal Summary")
+            out_nature = gr.Textbox(label="Distribution & Outlier Analysis", lines=10)
 
-        with gr.TabItem("3. Visual Mining & Nature"):
-            out_plot = gr.Image(label="15-Plot Distribution Dashboard")
-            out_nature = gr.Textbox(label="Nature of Data & Insights")
-            out_pca = gr.Label(label="Principal Component Variance")
+        with gr.TabItem("3. Visual Dashboard"):
+            out_corr = gr.Label(label="Correlation Inference")
+            out_img = gr.Image(label="EDA Dashboard (15 Plots)")
 
-        with gr.TabItem("4. Export"):
-            out_file = gr.File(label="Download Comprehensive Report")
+        with gr.TabItem("4. Download"):
+            out_file = gr.File(label="Final Report")
 
-    run_btn.click(
-        fn=perform_mining,
-        inputs=input_file,
-        outputs=[out_stats, out_desc, out_plot, out_nature, out_pca, out_file]
-    )
+    run_btn.click(perform_mining, u_file, [out_stats, out_desc, out_img, out_nature, out_corr, out_file])
 
 demo.launch()