Update app.py

app.py

@@ -1,6 +1,5 @@
 import pandas as pd
 import numpy as np
-from sklearn.ensemble import IsolationForest
 import gradio as gr
 import os
 import tempfile
@@ -68,16 +67,15 @@ def insert_reconciliation_to_salesforce(df, sf):
     return f"Inserted {inserted_count} records into Salesforce"
 
 def generate_suggestion(row):
-    if row['deviation'] > 15:
+    if row['deviation'] > 5:
         excess = row['used_quantity'] - row['planned_quantity']
         return f"Overuse Alert: Reduce future orders by {excess:.0f} units of {row['material_type']}."
-    elif row['deviation'] < -15:
+    elif row['deviation'] < -5:
         surplus = abs(row['planned_quantity'] - row['used_quantity'])
         return f"Surplus Detected: {surplus:.0f} units unused. Consider reducing future orders."
     return "Usage as planned. No action needed."
 
 def reconcile_materials(csv_file):
-    # Read CSV
     if isinstance(csv_file, str):
         df = pd.read_csv(csv_file)
     elif hasattr(csv_file, 'name'):
@@ -86,10 +84,8 @@ def reconcile_materials(csv_file):
         csv_file.seek(0)
         df = pd.read_csv(csv_file)
 
-    # Normalize column names: strip spaces, lowercase
     df.columns = df.columns.str.strip().str.lower().str.replace(' ', '_')
 
-    # Map common columns - all lowercase with underscores
     col_map = {
         'project_id': 'project_id',
         'material_type': 'material_type',
@@ -97,27 +93,25 @@ def reconcile_materials(csv_file):
         'received_quantity': 'received_quantity',
         'used_quantity': 'used_quantity',
     }
-    # Find columns from CSV that best match expected columns
+
     mapped_cols = {}
     for expected_col in col_map:
         for actual_col in df.columns:
             if actual_col == expected_col:
                 mapped_cols[expected_col] = actual_col
                 break
-    # Rename columns to expected names if present
+
     df.rename(columns=mapped_cols, inplace=True)
 
-    # Required columns
     required = ['material_type', 'planned_quantity', 'received_quantity', 'used_quantity']
     missing = [col for col in required if col not in df.columns]
     if missing:
         return None, f"Error: Missing required column(s): {', '.join(missing)}", None, None, None, None
 
-    # Convert quantities to numeric, fill missing with 0
     for col in ['planned_quantity', 'received_quantity', 'used_quantity']:
         df[col] = pd.to_numeric(df[col], errors='coerce').fillna(0)
 
-    # Compute balance and deviation
+    df['used_quantity'] = df[['used_quantity', 'received_quantity']].min(axis=1)
     df['balance_quantity'] = df['received_quantity'] - df['used_quantity']
     df['deviation'] = df.apply(
         lambda row: ((row['used_quantity'] - row['planned_quantity']) / row['planned_quantity']) * 100
@@ -125,25 +119,12 @@ def reconcile_materials(csv_file):
         axis=1
     )
 
-    # Anomaly detection
-    features = df[['planned_quantity', 'received_quantity', 'used_quantity', 'deviation']]
-    iso_forest = IsolationForest(contamination=0.1, random_state=42)
-    df['anomaly'] = iso_forest.fit_predict(features)
-
-    # Enforce anomaly for deviation beyond 15%
-    df.loc[df['deviation'].abs() > 15, 'anomaly'] = -1
-
-    # AI Suggestions & Status
+    df['anomaly'] = df['deviation'].apply(lambda d: -1 if abs(d) > 5 else 1)
     df['ai_suggestion'] = df.apply(generate_suggestion, axis=1)
-    df['reconciliation_status'] = df.apply(
-        lambda row: 'Flagged' if abs(row['deviation']) > 15 else 'Complete',
-        axis=1
-    )
+    df['reconciliation_status'] = df['deviation'].apply(lambda d: 'Flagged' if abs(d) > 5 else 'Complete')
 
-    # Insert into Salesforce (handle None for output)
     salesforce_result = insert_reconciliation_to_salesforce(df, sf)
 
-    # Prepare output text summary
     output_text = f"Material Reconciliation Results\n=============================\n\n"
     output_text += f"{salesforce_result}\n\nDetailed Records:\n"
     for i, row in df.iterrows():
@@ -161,12 +142,10 @@ def reconcile_materials(csv_file):
         output_text += f"  Reconciliation Status: {row['reconciliation_status']}\n"
         output_text += "-----------------------------\n"
 
-    # Save file
     with tempfile.NamedTemporaryFile(delete=False, suffix='.csv') as tmp:
         output_file = tmp.name
         df.to_csv(output_file, index=False)
 
-    # Charts
     bar_fig = px.bar(
         df, x='material_type', y='deviation',
         color='reconciliation_status',
@@ -186,7 +165,6 @@ def reconcile_materials(csv_file):
     ai_summary = "\n".join([f"{row['material_type']}: {row['ai_suggestion']}" for _, row in df.iterrows()])
     return output_file, output_text, df, bar_fig, pie_fig, ai_summary
 
-
 # Gradio UI
 with gr.Blocks(css='button:has(span:contains("Share via Link")) { display: none !important; }') as interface:
     gr.Markdown("# Material Reconciliation Dashboard")