Update app.py

app.py

@@ -727,7 +727,7 @@ def find_similar_movies(query_description, top_n=3):
     return movie_kb.find_similar_movies(query_description, top_n)
 
 # Original functions for analysis
-def predict_box_office(movie_description, similar_movies):
+def predict_box_office(movie_description, similar_movies, target_budget):
 
     import numpy as np
 
@@ -742,9 +742,13 @@ def predict_box_office(movie_description, similar_movies):
     for movie_info in similar_movies:
         movie = movie_info["movie"]
         sim_score = movie_info["similarity_score"]
+        if movie["budget"] > 0:
+          budget_ratio = target_budget / movie["budget"]
+        else:
+            budget_ratio = 1
         inflation_factor = get_inflation_adjustment(movie["year"])
         year_weight = get_year_weight(movie["year"])
-        adjusted_bo = movie["box_office"] * year_weight * inflation_factor
+        adjusted_bo = movie["box_office"] * year_weight * budget_ratio
         adjusted_box_offices.append(sim_score * adjusted_bo)
         total_weight += sim_score * year_weight
 
@@ -765,7 +769,7 @@ def predict_box_office(movie_description, similar_movies):
         genre_movies = [m for m in movie_knowledge_base if genre in m["genre"] and m["budget"] > 0]
         if not genre_movies:
             continue
-        success_rate = sum(1 for m in genre_movies if (m["box_office"] / m["budget"]) >= 2.5) / len(genre_movies)
+        success_rate = sum(1 for m in genre_movies if (m["box_office"] / m["budget"]) >= 3) / len(genre_movies)
         rois = [m["box_office"] / m["budget"] for m in genre_movies]
         roi_median = np.median(rois)
         revenues = [m["box_office"] for m in genre_movies]
@@ -789,7 +793,6 @@ def predict_box_office(movie_description, similar_movies):
 
     return simulations
 
-
 def predict_awards(movie_description, similar_movies):
     """Predict potential awards based on similar movies."""
     # Count awards in similar movies and recommend the most common ones
@@ -845,8 +848,17 @@ def get_similar_movies(movie_description: str):
 @log_function_tool(logger)
 def get_box_office_prediction(movie_description: str):
     """Predict the box office revenue for a movie based on its description."""
+    target_budget = None
+    import re
+    match = re.search(r"Budget:\s*:\s*(\d+)", movie_description, re.IGNORECASE)
+    if match:
+        target_budget = float(match.group(1))
+    else:
+        target_budget = 10000000
+    # print(f"the movie budget is {target_budget}")
+
     similar_movies = find_similar_movies(movie_description, top_n=3)
-    simulations = predict_box_office(movie_description, similar_movies)
+    simulations = predict_box_office(movie_description, similar_movies, target_budget)
     baseline_prediction = np.median(simulations)
     lower_bound = np.percentile(simulations, 25)
     upper_bound = np.percentile(simulations, 75)
@@ -857,8 +869,8 @@ def get_box_office_prediction(movie_description: str):
     else:
         avg_budget = 1
 
-    threshold_exceed = 3 * avg_budget
-    threshold_below = 2 * avg_budget
+    threshold_exceed = 3 * target_budget
+    threshold_below = 2 * target_budget
     prob_exceed = float(np.mean(simulations > threshold_exceed))
     prob_below = float(np.mean(simulations < threshold_below))
 
@@ -882,7 +894,6 @@ def get_box_office_prediction(movie_description: str):
         },
         "similar_movies": similar_movie_info
     }
-
 @function_tool
 @log_function_tool(logger)
 def get_award_predictions(movie_description: str):