Refactor load_overall_stats function in app.py: remove backlog data retrieval and associated UI elements to streamline data handling and improve performance

app.py

@@ -144,17 +144,8 @@ def load_overall_stats():
     roo_raw = raw_display.sort_values(by='Median', ascending=False)
 
     timestamp = raw_display['timestamp'].values[0]
-
-    collection = db["Range_Of_Outcomes_Backlog"] 
-    cursor = collection.find()
-
-    raw_display = pd.DataFrame(list(cursor))
-    raw_display = raw_display[['Player', 'Minutes Proj', 'Position', 'Team', 'Opp', 'Salary', 'Floor', 'Median', 'Ceiling', 'Top_finish', 'Top_5_finish', 'Top_10_finish', '20+%', '4x%', '5x%', '6x%', 'GPP%',
-                               'Own', 'Small_Own', 'Large_Own', 'Cash_Own', 'CPT_Own', 'LevX', 'ValX', 'site', 'version', 'slate', 'timestamp', 'Date']]
-    roo_backlog = raw_display.sort_values(by='Date', ascending=False)
-    roo_backlog = roo_backlog[roo_backlog['slate'] == 'Main Slate']
     
-    return dk_raw, fd_raw, dk_raw_sec, fd_raw_sec, roo_raw, timestamp, roo_backlog
+    return dk_raw, fd_raw, dk_raw_sec, fd_raw_sec, roo_raw, timestamp
 
 @st.cache_data(ttl = 60)
 def init_DK_lineups():  
@@ -204,7 +195,7 @@ def convert_df(array):
     array = pd.DataFrame(array, columns=column_names)
     return array.to_csv().encode('utf-8')
 
-dk_raw, fd_raw, dk_raw_sec, fd_raw_sec, roo_raw, timestamp, roo_backlog = load_overall_stats()
+dk_raw, fd_raw, dk_raw_sec, fd_raw_sec, roo_raw, timestamp = load_overall_stats()
 salary_dict = dict(zip(roo_raw.Player, roo_raw.Salary))
 
 try:
@@ -217,101 +208,6 @@ t_stamp = f"Last Update: " + str(timestamp) + f" CST"
 
 tab1, tab2 = st.tabs(['Range of Outcomes', 'Optimals'])
 
-with st.sidebar:
-    st.header("Quick Builder")
-    st.info("This is a quick hand building helper to give you some basic info about player combos and lineup feasibility")
-    sidebar_site = st.selectbox("What site are you running?", ('Draftkings', 'Fanduel'), key='sidebar_site')
-    sidebar_slate = st.selectbox("What slate are you running?", ('Main Slate', 'Secondary Slate'), key='sidebar_slate')
-
-    if sidebar_site == 'Draftkings':
-        roo_sample = roo_raw[roo_raw['slate'] == str(sidebar_slate)]
-        roo_sample = roo_sample[roo_sample['site'] == 'Draftkings']
-        roo_sample = roo_sample.sort_values(by='Own', ascending=False)
-        selected_pg = []
-        selected_sg = []
-        selected_sf = []
-        selected_pf = []
-        selected_c = []
-        selected_g = []
-        selected_f = []
-        selected_flex = []
-    elif sidebar_site == 'Fanduel':
-        roo_sample = roo_raw[roo_raw['slate'] == str(sidebar_slate)]
-        roo_sample = roo_sample[roo_sample['site'] == 'Fanduel']
-        roo_sample = roo_sample.sort_values(by='Own', ascending=False)
-        selected_pg1 = []
-        selected_pg2 = []
-        selected_sg1 = []
-        selected_sg2 = []
-        selected_sf1 = []
-        selected_sf2 = []
-        selected_pf1 = []
-        selected_pf2 = []
-        selected_c1 = []
-
-    # Get unique players by position from dk_roo_raw
-    pgs = roo_sample[roo_sample['Position'].str.contains('PG')]['Player'].unique()
-    sgs = roo_sample[roo_sample['Position'].str.contains('SG')]['Player'].unique()
-    sfs = roo_sample[roo_sample['Position'].str.contains('SF')]['Player'].unique()
-    pfs = roo_sample[roo_sample['Position'].str.contains('PF')]['Player'].unique()
-    centers = roo_sample[roo_sample['Position'].str.contains('C')]['Player'].unique()
-    guards = roo_sample[roo_sample['Position'].str.contains('G')]['Player'].unique()
-    forwards = roo_sample[roo_sample['Position'].str.contains('F')]['Player'].unique()
-    flex = roo_sample['Player'].unique()
-
-    if sidebar_site == 'Draftkings':
-        selected_pgs = st.multiselect('Select PG:', list(pgs), default=None, placeholder='Select PG', label_visibility='collapsed', key='pg1')
-        selected_sgs = st.multiselect('Select SG:', list(sgs), default=None, placeholder='Select SG', label_visibility='collapsed', key='sg1')
-        selected_sfs = st.multiselect('Select SF:', list(sfs), default=None, placeholder='Select SF', label_visibility='collapsed', key='sf1')
-        selected_pfs = st.multiselect('Select PF:', list(pfs), default=None, placeholder='Select PF', label_visibility='collapsed', key='pf1')
-        selected_cs = st.multiselect('Select C:', list(centers), default=None, placeholder='Select C', label_visibility='collapsed', key='c1')
-        selected_g = st.multiselect('Select G:', list(guards), default=None, placeholder='Select G', label_visibility='collapsed', key='g')
-        selected_f = st.multiselect('Select F:', list(forwards), default=None, placeholder='Select F', label_visibility='collapsed', key='f')
-        selected_flex = st.multiselect('Select Flex:', list(flex), default=None, placeholder='Select Flex', label_visibility='collapsed', key='flex')
-
-        # Combine all selected players
-        all_selected = selected_pgs + selected_sgs + selected_sfs + selected_pfs + selected_cs + selected_g + selected_f + selected_flex
-    
-    elif sidebar_site == 'Fanduel':
-        selected_pg1 = st.multiselect('Select PG1:', list(pgs), default=None, placeholder='Select PG1', label_visibility='collapsed', key='pg1')
-        selected_pg2 = st.multiselect('Select PG2:', list(pgs), default=None, placeholder='Select PG2', label_visibility='collapsed', key='pg2')
-        selected_sg1 = st.multiselect('Select SG1:', list(sgs), default=None, placeholder='Select SG1', label_visibility='collapsed', key='sg1')
-        selected_sg2 = st.multiselect('Select SG2:', list(sgs), default=None, placeholder='Select SG2', label_visibility='collapsed', key='sg2')
-        selected_sf1 = st.multiselect('Select SF1:', list(sfs), default=None, placeholder='Select SF1', label_visibility='collapsed', key='sf1')
-        selected_sf2 = st.multiselect('Select SF2:', list(sfs), default=None, placeholder='Select SF2', label_visibility='collapsed', key='sf2')
-        selected_pf1 = st.multiselect('Select PF1:', list(pfs), default=None, placeholder='Select PF1', label_visibility='collapsed', key='pf1')
-        selected_pf2 = st.multiselect('Select PF2:', list(pfs), default=None, placeholder='Select PF2', label_visibility='collapsed', key='pf2')
-        selected_c1 = st.multiselect('Select C1:', list(centers), default=None, placeholder='Select C1', label_visibility='collapsed', key='c1')
-
-        # Combine all selected players
-        all_selected = selected_pg1 + selected_pg2 + selected_sg1 + selected_sg2 + selected_sf1 + selected_sf2 + selected_pf1 + selected_pf2 + selected_c1
-
-    if all_selected:
-        # Get stats for selected players
-        selected_stats = roo_sample[roo_sample['Player'].isin(all_selected)]
-        
-        # Calculate sums
-        salary_sum = selected_stats['Salary'].sum()
-        median_sum = selected_stats['Median'].sum()
-        own_sum = selected_stats['Own'].sum()
-        levx_sum = selected_stats['LevX'].sum()
-
-        # Display sums
-        st.write('---')
-        if sidebar_site == 'Draftkings':
-            if salary_sum > 50000:
-                st.warning(f'Total Salary: ${salary_sum:.2f} exceeds limit of $50,000')
-            else:
-                st.write(f'Total Salary: ${salary_sum:.2f}')
-        elif sidebar_site == 'Fanduel':
-            if salary_sum > 60000:
-                st.warning(f'Total Salary: ${salary_sum:.2f} exceeds limit of $60,000')
-            else:
-                st.write(f'Total Salary: ${salary_sum:.2f}')
-        st.write(f'Total Median: {median_sum:.2f}')
-        st.write(f'Total Ownership: {own_sum:.2f}%')
-        st.write(f'Total LevX: {levx_sum:.2f}')
-
 with tab1:
     
     with st.expander("Info and Filters"):
@@ -325,7 +221,7 @@ with tab1:
             with col2:
                 if st.button("Load/Reset Data", key='reset1'):
                     st.cache_data.clear()
-                    dk_raw, fd_raw, dk_raw_sec, fd_raw_sec, roo_raw, timestamp, roo_backlog = load_overall_stats()
+                    dk_raw, fd_raw, dk_raw_sec, fd_raw_sec, roo_raw, timestamp = load_overall_stats()
                     id_dict = dict(zip(roo_raw.Player, roo_raw.player_ID))
                     dk_lineups = init_DK_lineups()
                     fd_lineups = init_FD_lineups()
@@ -341,10 +237,8 @@ with tab1:
             # Process site selection
             if site_var2 == 'Draftkings':
                 site_baselines = roo_raw[roo_raw['site'] == 'Draftkings']
-                site_backlog = roo_backlog[roo_backlog['site'] == 'Draftkings']
             elif site_var2 == 'Fanduel':
                 site_baselines = roo_raw[roo_raw['site'] == 'Fanduel']
-                site_backlog = roo_backlog[roo_backlog['site'] == 'Fanduel']
         with col3:
             slate_split = st.radio("Slate Type", ('Main Slate', 'Secondary', 'Backlog'), key='slate_split')
 
@@ -352,21 +246,6 @@ with tab1:
                 raw_baselines = site_baselines[site_baselines['slate'] == 'Main Slate']
             elif slate_split == 'Secondary':
                 raw_baselines = site_baselines[site_baselines['slate'] == 'Secondary Slate']
-            elif slate_split == 'Backlog':
-                raw_baselines = site_backlog
-                # Third row - backlog options
-                col1, col2 = st.columns(2)
-                with col1:
-                    view_all = st.checkbox("View all dates?", key='view_all')
-                with col2:
-                    if not view_all:
-                        date_var2 = st.date_input("Select date", key='date_var2')
-                
-                if view_all:
-                    raw_baselines = raw_baselines.sort_values(by=['Median', 'Date'], ascending=[False, False])
-                else:
-                    raw_baselines = raw_baselines[raw_baselines['Date'] == date_var2.strftime('%m-%d-%Y')]
-                    raw_baselines = raw_baselines.sort_values(by='Median', ascending=False)
         
         with col4:
             split_var2 = st.radio("Slate Range", ('Full Slate Run', 'Specific Games'), key='split_var2')