Update app.py

app.py

@@ -4,7 +4,6 @@ from fastapi import FastAPI, Request
 from fastapi.responses import JSONResponse
 import uvicorn
 from sklearn.linear_model import LinearRegression
-from sklearn.model_selection import train_test_split
 import base64
 import os
 from datetime import datetime
@@ -18,9 +17,12 @@ import matplotlib.pyplot as plt
 import seaborn as sns
 import numpy as np
 
+# Import Hugging Face transformers pipeline for text generation
+from transformers import pipeline
+
 weather_map = {"Cloudy": 0, "Rainy": 1, "Sunny": 2}
 
-# Load and preprocess training data
+# Load and preprocess data
 print("Loading and preprocessing data...")
 try:
     if not os.path.exists("new_delay_data.csv"):
@@ -56,7 +58,7 @@ except Exception as e:
     print(f"Error preparing features: {e}")
     raise
 
-# Train model
+# Train Linear Regression model
 print("Training model...")
 try:
     model = LinearRegression()
@@ -66,7 +68,35 @@ except Exception as e:
     print(f"Error training model: {e}")
     raise
 
-# Function to generate heatmap and return path
+# Load GPT-2 text generation pipeline for AI insights and migration plans
+print("Loading AI text generation model...")
+try:
+    text_generator = pipeline("text-generation", model="gpt2")
+    print("Text generation model loaded successfully.")
+except Exception as e:
+    print(f"Failed to load text generation model: {e}")
+    raise
+
+# Generate AI insight and migration plan dynamically
+def generate_ai_insight(phase, weather, absentee_pct, delay_log, prediction):
+    prompt = (
+        f"Project phase: {phase}. Weather: {weather}. "
+        f"Absentee percentage: {absentee_pct}%. Previous delay log: {delay_log}. "
+        f"Predicted delay: {prediction}%. "
+        "Provide a concise AI-generated insight about delay risks and a practical migration plan "
+        "to reduce these delays and improve project efficiency."
+    )
+    try:
+        result = text_generator(prompt, max_length=150, num_return_sequences=1)
+        generated_text = result[0]['generated_text']
+        # Strip out prompt to get only generated part
+        insight = generated_text[len(prompt):].strip()
+        return insight
+    except Exception as e:
+        print(f"AI insight generation failed: {e}")
+        return "AI insight generation failed. Please check logs."
+
+# Heatmap generation function (unchanged)
 def generate_heatmap(phase, weather, model):
     print("Generating heatmap...")
     try:
@@ -101,16 +131,14 @@ def generate_heatmap(phase, weather, model):
         print(f"Heatmap generation failed: {e}")
         return None
 
-# Function to generate simple PDF with heatmap and return base64-encoded string
+# PDF generation function (unchanged except uses AI-generated insight)
 def generate_pdf_report(phase, weather, absentee_pct, delay_log, prediction, risk, insight):
     print("Generating PDF report...")
     try:
         buffer = BytesIO()
         c = canvas.Canvas(buffer, pagesize=letter)
         
-        # Register DejaVuSans font to support Unicode emojis
         try:
-            # Assuming DejaVuSans.ttf is available in the project directory
             pdfmetrics.registerFont(TTFont('DejaVuSans', 'DejaVuSans.ttf'))
             c.setFont("DejaVuSans", 12)
             flag_indicator = " 🚩" if prediction >= 75 else ""
@@ -123,7 +151,7 @@ def generate_pdf_report(phase, weather, absentee_pct, delay_log, prediction, ris
         c.drawString(100, 730, f"Generated on: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
         
         y_position = 700
-        max_width = 400  # max width in points for wrapping text
+        max_width = 400
         
         details = [
             f"Phase: {phase}",
@@ -132,23 +160,20 @@ def generate_pdf_report(phase, weather, absentee_pct, delay_log, prediction, ris
             f"Previous Delay Log: {delay_log}",
             f"Predicted Delay: {prediction}%{flag_indicator}",
             f"Risk Level: {risk}",
-            "AI Insight:"
+            "AI Insight & Migration Plan:"
         ]
         
-        # Wrap and draw each line properly
         for line in details:
             lines = simpleSplit(line, 'DejaVuSans' if 'DejaVuSans' in pdfmetrics.getRegisteredFontNames() else 'Helvetica', 12, max_width)
             for wrapped_line in lines:
                 c.drawString(100, y_position, wrapped_line)
                 y_position -= 16
         
-        # Wrap and draw insight (which may be long)
         insight_lines = simpleSplit(insight, 'DejaVuSans' if 'DejaVuSans' in pdfmetrics.getRegisteredFontNames() else 'Helvetica', 12, max_width)
         for wrapped_line in insight_lines:
             c.drawString(100, y_position, wrapped_line)
             y_position -= 16
         
-        # Add heatmap
         heatmap_path = generate_heatmap(phase, weather, model)
         if heatmap_path and os.path.exists(heatmap_path):
             c.drawString(100, y_position - 20, "Delay Prediction Heatmap:")
@@ -175,7 +200,7 @@ def generate_pdf_report(phase, weather, absentee_pct, delay_log, prediction, ris
         print(f"PDF generation failed: {e}")
         return None, None, None
 
-# Main prediction function
+# Main prediction logic
 def predict_delay(phase, weather, absentee_pct, delay_log):
     print(f"Predicting delay for Phase: {phase}, Weather: {weather}, Absentee: {absentee_pct}, Delay Log: {delay_log}")
     try:
@@ -194,60 +219,14 @@ def predict_delay(phase, weather, absentee_pct, delay_log):
         prediction = model.predict(input_data)[0]
         prediction = round(prediction, 2)
 
-        # Tailored AI Insights
         if prediction >= 75:
             risk = "High Risk"
-            insight = f"High delay risk ({prediction}%) in {phase} phase under {weather} conditions. "
-            if absentee_pct > 30:
-                insight += f"High absenteeism ({absentee_pct}%) is a major factor. Hire temporary workers or offer overtime incentives. "
-            else:
-                insight += f"Absenteeism ({absentee_pct}%) is moderate; ensure key staff are present for critical {phase} tasks. "
-            if delay_log > 5:
-                insight += f"Significant past delays ({delay_log}) detected; conduct a root cause analysis to address bottlenecks. "
-            else:
-                insight += f"Past delays ({delay_log}) are manageable; review task dependencies to prevent escalation. "
-            if weather == "Rainy":
-                insight += "Rainy weather increases risk; use protective coverings or shift to indoor tasks."
-            elif weather == "Cloudy":
-                insight += "Cloudy weather may slow progress; monitor conditions and prepare for potential rain."
-            else:
-                insight += "Sunny weather is optimal; maximize outdoor work to reduce delays."
         elif prediction >= 50:
             risk = "Moderate Risk"
-            insight = f"Moderate delay risk ({prediction}%) in {phase} phase under {weather} conditions. "
-            if absentee_pct > 30:
-                insight += f"High absenteeism ({absentee_pct}%) needs attention; consider cross-training staff to cover gaps. "
-            elif absentee_pct < 10:
-                insight += f"Low absenteeism ({absentee_pct}%) is good; maintain attendance with morale-boosting measures. "
-            else:
-                insight += f"Moderate absenteeism ({absentee_pct}%) suggests reviewing workforce allocation for {phase} tasks. "
-            if delay_log > 5:
-                insight += f"Past delays ({delay_log}) indicate inefficiencies; streamline workflows in {phase}. "
-            else:
-                insight += f"Past delays ({delay_log}) are low; ensure timely material delivery to maintain progress. "
-            if weather == "Rainy":
-                insight += "Rainy weather may disrupt work; schedule flexible tasks and secure equipment."
-            elif weather == "Cloudy":
-                insight += "Cloudy weather is manageable; keep weather monitoring active."
-            else:
-                insight += "Sunny weather supports progress; optimize daily schedules to leverage good conditions."
         else:
             risk = "Low Risk"
-            insight = f"Low delay risk ({prediction}%) in {phase} phase under {weather} conditions. "
-            if absentee_pct > 30:
-                insight += f"Despite low risk, high absenteeism ({absentee_pct}%) could escalate; monitor attendance closely. "
-            else:
-                insight += f"Absenteeism ({absentee_pct}%) is under control; sustain with regular team check-ins. "
-            if delay_log > 5:
-                insight += f"Past delays ({delay_log}) are notable; maintain vigilance to prevent recurrence in {phase}. "
-            else:
-                insight += f"Minimal past delays ({delay_log}); continue efficient task management in {phase}. "
-            if weather == "Rainy":
-                insight += "Rainy weather could pose minor risks; have contingency plans ready."
-            elif weather == "Cloudy":
-                insight += "Cloudy weather is unlikely to cause issues; maintain standard operations."
-            else:
-                insight += "Sunny weather is ideal; capitalize on it to stay ahead of schedule."
+
+        insight = generate_ai_insight(phase, weather, absentee_pct, delay_log, prediction)
 
         pdf_base64, pdf_path, heatmap_path = generate_pdf_report(phase, weather, absentee_pct, delay_log, prediction, risk, insight)
 
@@ -256,7 +235,7 @@ def predict_delay(phase, weather, absentee_pct, delay_log):
         print(f"Prediction error: {e}")
         return None, None, f"Error: {e}", None, None, None
 
-# FastAPI for Salesforce
+# FastAPI app
 api_app = FastAPI()
 
 @api_app.post("/predict")
@@ -287,11 +266,11 @@ async def predict_from_salesforce(request: Request):
         print(f"API error: {e}")
         return JSONResponse(status_code=500, content={"status": "error", "message": str(e)})
 
-# Gradio UI for manual testing
+# Gradio UI
 try:
     print("Setting up Gradio UI...")
     with gr.Blocks() as demo:
-        gr.Markdown("## 🏗️ Delay Predictor")
+        gr.Markdown("## 🏗️ Delay Predictor with AI Insights")
         with gr.Row():
             phase_input = gr.Textbox(label="Phase (Framing/Foundation/Finishing)", value="Framing")
             weather_input = gr.Textbox(label="Weather (Sunny/Rainy/Cloudy)", value="Sunny")
@@ -309,7 +288,7 @@ try:
             flag = " 🚩" if prediction >= 75 else ""
             return (f"Predicted Delay: {prediction}%{flag}\n"
                     f"Risk Level: {risk}\n"
-                    f"Insight: {insight}\n"
+                    f"Insight & Migration Plan:\n{insight}\n\n"
                     f"PDF Report: {'Saved locally at ' + pdf_path if pdf_path else 'Failed to generate'}\n"
                     f"Heatmap: {'Saved locally at ' + heatmap_path if heatmap_path else 'Failed to generate'}\n"
                     f"PDF Base64: {'Generated' if pdf_base64 else 'Not generated'}")
@@ -324,7 +303,7 @@ except Exception as e:
     print(f"Error setting up Gradio UI: {e}")
     raise
 
-# Mount Gradio inside FastAPI
+# Mount Gradio app inside FastAPI
 try:
     print("Mounting Gradio app...")
     app = gr.mount_gradio_app(api_app, demo, path="/")
@@ -333,11 +312,10 @@ except Exception as e:
     print(f"Error mounting Gradio app: {e}")
     raise
 
-# Run locally
 if __name__ == "__main__":
     print("Starting server on http://0.0.0.0:7860...")
     try:
         uvicorn.run(app, host="0.0.0.0", port=7860)
     except Exception as e:
         print(f"Server failed to start: {e}")
-        raise
+        raise