Clean dashboard script to avoid regenerating figures

Author: pmichaillat

content/dashboard/dashboard.py

@@ -140,6 +140,7 @@ def make_plot(df, y_column, title, filename, y_label, x_min=None, x_max=None, y_
         html_file_path,
         include_plotlyjs='cdn',
         full_html=False,
+        div_id=filename, 
         config={
             "displaylogo": False,
             "modeBarButtonsToRemove": [
@@ -159,36 +160,40 @@ def make_plot(df, y_column, title, filename, y_label, x_min=None, x_max=None, y_
 
     print(f"Successfully wrote HTML: {html_file_path}")
 
-
 # Fetch raw data from FRED
+# Fetch unemployment level
 u = fred.get_series("UNEMPLOY")
 if u is not None and not u.empty:
-    print(f"UNEMPLOY data fetched successfully. Shape: {u.shape}, Last date: {u.index.max()}")
+    print(f"UNEMPLOY data fetched successfully: shape = {u.shape}")
     u.index = pd.to_datetime(u.index)
     latest_u_date = u.index.max()
     latest_u_value = u.iloc[-1]
-    print(f"UNEMPLOY raw latest point: Date='{latest_u_date.strftime('%Y-%m-%d')}', Value='{latest_u_value}'")
+    previous_u_value = u.iloc[-2]
+    print(f"UNEMPLOY raw latest point: date = {latest_u_date.strftime('%Y-%m-%d')}, value = {latest_u_value}, previous value = {previous_u_value}")
 else:
     print("Failed to fetch UNEMPLOY data or data is empty.")
 
+# Fetch vacancy level
 v = fred.get_series("JTSJOL")
 if v is not None and not v.empty:
-    print(f"JTSJOL data fetched successfully. Shape: {v.shape}, Last date: {v.index.max()}")
+    print(f"JTSJOL data fetched successfully: shape = {v.shape}")
     v.index = pd.to_datetime(v.index)
     latest_v_date = v.index.max()
     latest_v_value = v.iloc[-1]
     previous_v_value = v.iloc[-2]
-    print(f"JTSJOL raw latest point: Date='{latest_v_date.strftime('%Y-%m-%d')}', Latest value='{latest_v_value}', Previous value='{previous_v_value}'")
+    print(f"JTSJOL raw latest point: date = {latest_v_date.strftime('%Y-%m-%d')}, value = {latest_v_value}, previous value = {previous_v_value}")
 else:
     print("Failed to fetch JTSJOL data or data is empty.")
 
+# Fetch labor force level
 lf = fred.get_series("CLF16OV")
 if lf is not None and not lf.empty:
-    print(f"CLF16OV data fetched successfully. Shape: {lf.shape}, Last date: {lf.index.max()}")
+    print(f"CLF16OV data fetched successfully: shape = {lf.shape}")
     lf.index = pd.to_datetime(lf.index)
     latest_lf_date = lf.index.max()
     latest_lf_value = lf.iloc[-1]
-    print(f"CLF16OV raw latest point: Date='{latest_lf_date.strftime('%Y-%m-%d')}', Value='{latest_lf_value}'")
+    previous_lf_value = lf.iloc[-2]
+    print(f"CLF16OV raw latest point: date = {latest_lf_date.strftime('%Y-%m-%d')}, value = {latest_lf_value}, previous value = {previous_lf_value}")
 
 else:
     print("Failed to fetch CLF16OV data or data is empty.")
@@ -253,7 +258,7 @@ def make_plot(df, y_column, title, filename, y_label, x_min=None, x_max=None, y_
     latest_processed_u_date = df.index.max()
     latest_processed_u_value = df['data'].iloc[-1]
     previous_processed_u_value = df['data'].iloc[-2]
-    print(f"Unemployment rate latest processed point for CSV: Date='{latest_processed_u_date.strftime('%Y-%m-%d')}', Value='{latest_processed_u_value:.4f}', Other value='{previous_processed_u_value:.4f}'")
+    print(f"Unemployment rate latest processed point for CSV: date = {latest_processed_u_date.strftime('%Y-%m-%d')}, value = {latest_processed_u_value:.4f}, previous value = {previous_processed_u_value:.4f}")
 else:
     print("'df' is EMPTY before writing CSV.")
 
@@ -286,7 +291,7 @@ def make_plot(df, y_column, title, filename, y_label, x_min=None, x_max=None, y_
     latest_processed_v_date = df.index.max()
     latest_processed_v_value = df['data'].iloc[-1]
     previous_processed_v_value = df['data'].iloc[-2]
-    print(f"Vacancy rate latest processed point for CSV: Date='{latest_processed_v_date.strftime('%Y-%m-%d')}', Value='{latest_processed_v_value:.4f}', Other value='{previous_processed_v_value:.4f}'")
+    print(f"Vacancy rate latest processed point for CSV: date = {latest_processed_v_date.strftime('%Y-%m-%d')}, value = {latest_processed_v_value:.4f}, previous value = {previous_processed_v_value:.4f}")
 else:
     print("'df' is EMPTY before writing CSV.")
 
@@ -314,12 +319,8 @@ def make_plot(df, y_column, title, filename, y_label, x_min=None, x_max=None, y_
 # Immediately after writing, read back and print tail in Python
 try:
     df_read_back = pd.read_csv(csv_path_absolute, index_col="Date", parse_dates=True)
-    print("Last 5 rows of vacancy_rate.csv as read by Python immediately after write:")
-    print(df_read_back.tail(5).to_string())
-    latest_read_back_v_date = df_read_back.index.max()
-    latest_read_back_v_value = df_read_back["Vacancy rate (%)"].iloc[-1]
-    previous_read_back_v_value = df_read_back["Vacancy rate (%)"].iloc[-2]
-    print(f"Vacancy rate latest readback point: Date='{latest_read_back_v_date.strftime('%Y-%m-%d')}', Value='{latest_read_back_v_value:.4f}', Other value='{previous_read_back_v_value:.4f}'")
+    print("Last 3 rows of vacancy_rate.csv as read by Python immediately after write:")
+    print(df_read_back.tail(3).to_string())
 
 except Exception as e:
     print(f"Error reading back {csv_path_absolute} in Python: {e}")
@@ -441,6 +442,7 @@ def make_plot(df, y_column, title, filename, y_label, x_min=None, x_max=None, y_
     beveridge_curve_html_path,
     include_plotlyjs='cdn',
     full_html=False,
+    div_id="beveridge_curve", 
     config={
         "displaylogo": False,
         "modeBarButtonsToRemove": [

content/papers/16.md

@@ -2,7 +2,7 @@
 title: "Has the Recession Started?" 
 date: 2025-05-09
 url: /16/
-tags: ["Beveridge curve", "business cycles", "job vacancies", "Michez rule", "nowcasting", "recession indicator", "recession probability", "recession threshold", "Sahm rule", "unemployment"]
+tags: ["Beveridge curve", "business cycles", "job vacancies", "Michez rule", "nowcasting", "recession indicator", "recession probability", "recession threshold", "Sahm rule"]
 author: ["Pascal Michaillat","Emmanuel Saez"]
 description: "This paper builds the Michez rule, a fast and robust method to detect US recessions from data on unemployment and job vacancies. Published in OBES, 2025."
 summary: "This paper develops the Michez rule, a fast and robust method to detect US recessions using data on unemployment and job vacancies. In December 2024, the rule gives a recession probability of 27%; this recession risk was first detected in March 2024."

content/papers/17.md

@@ -2,7 +2,7 @@
 title: "Early and Accurate Recession Detection Using Classifiers on the Anticipation-Precision Frontier" 
 date: 2025-06-11
 url: /17/
-tags: ["backtesting", "Beveridge curve", "business cycles", "classifier ensemble", "job vacancies", "nowcasting", "perfect classifiers", "recession probability", "unemployment"]
+tags: ["backtesting", "Beveridge curve", "business cycles", "classifier ensemble", "job vacancies", "nowcasting", "perfect classifiers", "recession probability"]
 author: ["Pascal Michaillat"]
 description: "This paper detects US recessions in real time from unemployment and vacancy data by combining perfect classifiers on the anticipation-precision frontier."
 summary: "This paper develops a method for detecting US recessions in real time using unemployment and vacancy data. The method combines perfect recession classifiers on the anticipation-precision frontier. Backtesting confirms that recessions are detected early and accurately."