Update app.py

app.py

@@ -69,14 +69,20 @@ def determine_frequency(df):
     df = df.drop_duplicates(subset='ds')
     df = df.set_index('ds')
     
-    # Create a complete date range
-    full_range = pd.date_range(start=df.index.min(), end=df.index.max())
+    # # Create a complete date range
+    # full_range = pd.date_range(start=df.index.min(), end=df.index.max(),freq=freq)
     
-    # Reindex the DataFrame to this full date range
-    df_full = df.reindex(full_range)
+    # # Reindex the DataFrame to this full date range
+    # df_full = df.reindex(full_range)
     
     # Infer the frequency
-    freq = pd.infer_freq(df_full.index)
+    # freq = pd.infer_freq(df_full.index)
+
+    freq = pd.infer_freq(df.index)
+    if not freq:
+        st.warning('The forecast will use default Daily forecast due to date inconsistency. Please check your data.',icon="⚠️")
+        freq = 'D'
+        
     return freq
 
 def plot_forecasts(forecast_df, train_df, title):
@@ -375,14 +381,13 @@ def timegpt_fcst():
                 freq=freq,
                 level=[90]
             )
-            st.write(forecast_df)
 
-            st.pyplot(nixtla_client.plot(
-                df,
-                forecast_df,
-                level=[90],
-                max_insample_length=365
-            ))
+    st.pyplot(nixtla_client.plot(
+        df,
+        forecast_df,
+        level=[90],
+        max_insample_length=365
+    ))
 
 
 
@@ -431,10 +436,10 @@ def timegpt_anom():
                 level=90
             )
 
-            nixtla_client.plot(
-                df,
-                anom_df
-            )
+    st.pyplot(nixtla_client.plot(
+        df,
+        anom_df
+    ))