Forest Service Accessions and Separations Analysis

Author

Abigail Haddad

Published

July 12, 2025

# Import required libraries
import pandas as pd
import numpy as np
from pathlib import Path
import matplotlib.pyplot as plt
import seaborn as sns
from great_tables import GT, style, loc

# Configuration
PARQUET_DIR = "../../separations_accessions/parquet/"
FOREST_SERVICE_CODE = "AG11"

# Define color scales for heatmaps
ACCESSIONS_COLOR = "#1f77b4"  # Blue for accessions
SEPARATIONS_COLOR = "#ff7f0e"  # Orange for separations
NET_POSITIVE_COLOR = "#2ca02c"  # Green for positive net
NET_NEGATIVE_COLOR = "#d62728"  # Red for negative net

Data Loading

Show code

# Load accessions data
accessions_files = [
    "fedscope_accessions_FY2005-2009.parquet",
    "fedscope_accessions_FY2010-2014.parquet", 
    "fedscope_accessions_FY2015-2019.parquet",
    "fedscope_accessions_FY2020-2024.parquet",
    "fedscope_accessions_March_2025.parquet"
]

# Load separations data
separations_files = [
    "fedscope_separations_FY2005-2009.parquet",
    "fedscope_separations_FY2010-2014.parquet",
    "fedscope_separations_FY2015-2019.parquet", 
    "fedscope_separations_FY2020-2024.parquet",
    "fedscope_separations_March_2025.parquet"
]

# Load all accessions data
accessions_data = []
for file in accessions_files:
    try:
        df = pd.read_parquet(PARQUET_DIR + file)
        # Handle both uppercase and lowercase column names
        agysub_col = 'AGYSUB' if 'AGYSUB' in df.columns else 'agysub'
        # Filter for Forest Service
        df_forest = df[df[agysub_col] == FOREST_SERVICE_CODE].copy()
        if not df_forest.empty:
            accessions_data.append(df_forest)
            print(f"Loaded {len(df_forest)} Forest Service accessions records from {file}")
    except Exception as e:
        print(f"Error loading {file}: {e}")

# Load all separations data  
separations_data = []
for file in separations_files:
    try:
        df = pd.read_parquet(PARQUET_DIR + file)
        # Filter for Forest Service
        df_forest = df[df['agysub'] == FOREST_SERVICE_CODE].copy()
        if not df_forest.empty:
            separations_data.append(df_forest)
            print(f"Loaded {len(df_forest)} Forest Service separations records from {file}")
    except Exception as e:
        print(f"Error loading {file}: {e}")

# Combine all data
all_accessions = pd.concat(accessions_data, ignore_index=True) if accessions_data else pd.DataFrame()
all_separations = pd.concat(separations_data, ignore_index=True) if separations_data else pd.DataFrame()

print(f"\nTotal Forest Service accessions records: {len(all_accessions):,}")
print(f"Total Forest Service separations records: {len(all_separations):,}")

Loaded 74461 Forest Service accessions records from fedscope_accessions_FY2005-2009.parquet
Loaded 69871 Forest Service accessions records from fedscope_accessions_FY2010-2014.parquet
Loaded 67422 Forest Service accessions records from fedscope_accessions_FY2015-2019.parquet
Loaded 49603 Forest Service accessions records from fedscope_accessions_FY2020-2024.parquet
Loaded 8031 Forest Service accessions records from fedscope_accessions_March_2025.parquet
Loaded 77754 Forest Service separations records from fedscope_separations_FY2005-2009.parquet
Loaded 71494 Forest Service separations records from fedscope_separations_FY2010-2014.parquet
Loaded 68659 Forest Service separations records from fedscope_separations_FY2015-2019.parquet
Loaded 45472 Forest Service separations records from fedscope_separations_FY2020-2024.parquet
Loaded 8249 Forest Service separations records from fedscope_separations_March_2025.parquet

Total Forest Service accessions records: 269,388
Total Forest Service separations records: 271,628

Monthly Aggregation

Show code

def aggregate_by_month(df, data_type):
    """Aggregate data by month (EFDATE field)."""
    if df.empty:
        return pd.DataFrame()
    
    # Convert EFDATE to string and extract year/month
    df['efdate_str'] = df['efdate'].astype(str)
    df['year'] = df['efdate_str'].str[:4].astype(int)
    df['month'] = df['efdate_str'].str[4:6].astype(int)
    
    # Group by year and month
    monthly = df.groupby(['year', 'month']).size().reset_index(name='count')
    monthly['data_type'] = data_type
    
    return monthly

# Aggregate accessions and separations by month
monthly_accessions = aggregate_by_month(all_accessions, 'accessions')
monthly_separations = aggregate_by_month(all_separations, 'separations')

# Merge to calculate net change
if not monthly_accessions.empty and not monthly_separations.empty:
    monthly_combined = pd.merge(
        monthly_accessions[['year', 'month', 'count']],
        monthly_separations[['year', 'month', 'count']],
        on=['year', 'month'],
        how='outer',
        suffixes=('_acc', '_sep')
    ).fillna(0)
    
    monthly_combined['accessions'] = monthly_combined['count_acc'].astype(int)
    monthly_combined['separations'] = monthly_combined['count_sep'].astype(int)
    monthly_combined['net_change'] = monthly_combined['accessions'] - monthly_combined['separations']
    monthly_combined = monthly_combined.drop(['count_acc', 'count_sep'], axis=1)
else:
    monthly_combined = pd.DataFrame()

print(f"Monthly data points: {len(monthly_combined)}")

Monthly data points: 240

Accessions Heatmap

Show code

if not monthly_combined.empty:
    # Create pivot table for accessions
    accessions_pivot = monthly_combined.pivot(index='month', columns='year', values='accessions')
    
    # Get most recent 10 years for main display
    all_years = sorted(accessions_pivot.columns)
    recent_years = all_years[-10:]  # Last 10 years
    recent_accessions = accessions_pivot[recent_years]
    
    # Create month labels
    month_labels = ['Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun', 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec']
    recent_accessions.index = [month_labels[i-1] for i in recent_accessions.index]
    
    # Create GT heatmap for accessions (recent years)
    accessions_df = recent_accessions.reset_index()
    accessions_df.columns = ['Month'] + [str(col) for col in recent_accessions.columns]
    
    gt_accessions = (
        GT(accessions_df)
        .tab_header(title=f"Forest Service Monthly Accessions ({recent_years[0]}-{recent_years[-1]})")
        .fmt_number(columns=list(accessions_df.columns[1:]), decimals=0, use_seps=True)
        .data_color(
            columns=list(accessions_df.columns[1:]),
            palette=["white", ACCESSIONS_COLOR],
            domain=[0, accessions_pivot.max().max()],
            na_color="lightgray"
        )
    )
    
    gt_accessions.show()

Forest Service Monthly Accessions (2016-2025)
Month	2016	2017	2018	2019	2020	2021	2022	2023	2024	2025
Jan	350	784	238	83	272	351	285	631		123
Feb	300	193	247	233	293	309	441	436		111
Mar	738	422	556	944	1,377	911	1,062	1,125		502
Apr	2,163	4,480	4,088	3,720	2,644	3,035	2,984	2,884	1,949
May	7,735	5,245	5,373	4,987	5,556	4,566	3,605	3,225	2,910
Jun	1,336	1,354	1,252	1,406	1,450	1,144	1,408	2,046	1,744
Jul	488	371	398	341	402	288	899	788	221
Aug	271	235	317	331	503	260	403	415	151
Sep	198	99	269	251	234	229	304	383	82
Oct	300	151	156	234	215	213	393		102
Nov	139	107	156	178	143	154	267		82
Dec	186	97	88	153	143	116	246		54

Separations Heatmap

Show code

if not monthly_combined.empty:
    # Create pivot table for separations
    separations_pivot = monthly_combined.pivot(index='month', columns='year', values='separations')
    
    # Get most recent 10 years for main display
    recent_separations = separations_pivot[recent_years]
    
    # Create month labels
    recent_separations.index = [month_labels[i-1] for i in recent_separations.index]
    
    # Create GT heatmap for separations (recent years)
    separations_df = recent_separations.reset_index()
    separations_df.columns = ['Month'] + [str(col) for col in recent_separations.columns]
    
    gt_separations = (
        GT(separations_df)
        .tab_header(title=f"Forest Service Monthly Separations ({recent_years[0]}-{recent_years[-1]})")
        .fmt_number(columns=list(separations_df.columns[1:]), decimals=0, use_seps=True)
        .data_color(
            columns=list(separations_df.columns[1:]),
            palette=["white", SEPARATIONS_COLOR],
            domain=[0, separations_pivot.max().max()],
            na_color="lightgray"
        )
    )
    
    gt_separations.show()

Forest Service Monthly Separations (2016-2025)
Month	2016	2017	2018	2019	2020	2021	2022	2023	2024	2025
Jan	517	533	449	382	454	504	374	341		378
Feb	271	258	264	367	326	245	351	337		279
Mar	305	314	470	424	299	350	378	401		434
Apr	455	386	394	354	306	412	414	362	439
May	369	366	372	369	290	376	399	288	372
Jun	366	343	364	332	310	357	340	375	352
Jul	385	391	345	374	440	602	509	381	369
Aug	1,622	1,550	1,512	1,539	1,359	1,362	1,203	1,255	1,043
Sep	2,153	2,648	2,446	2,278	1,623	1,703	1,558	1,248	1,086
Oct	4,403	3,836	3,554	3,169	2,647	2,867	2,147		1,744
Nov	2,267	2,449	2,394	2,341	2,449	2,171	1,746		1,219
Dec	902	772	785	1,005	1,016	1,079	1,003		534

Net Change Heatmap

Show code

if not monthly_combined.empty:
    # Create pivot table for net change
    net_pivot = monthly_combined.pivot(index='month', columns='year', values='net_change')
    
    # Get most recent 10 years for main display
    recent_net = net_pivot[recent_years]
    
    # Create month labels
    recent_net.index = [month_labels[i-1] for i in recent_net.index]
    
    # Get max absolute value for symmetric scale (from all data)
    max_abs_val = max(abs(net_pivot.min().min()), abs(net_pivot.max().max()))
    
    # Create GT heatmap for net change with diverging colors (recent years)
    net_df = recent_net.reset_index()
    net_df.columns = ['Month'] + [str(col) for col in recent_net.columns]
    
    gt_net = (
        GT(net_df)
        .tab_header(title=f"Forest Service Monthly Net Change ({recent_years[0]}-{recent_years[-1]})")
        .fmt_number(columns=list(net_df.columns[1:]), decimals=0, use_seps=True)
        .data_color(
            columns=list(net_df.columns[1:]),
            palette=[NET_NEGATIVE_COLOR, "white", NET_POSITIVE_COLOR],
            domain=[-max_abs_val, max_abs_val],
            na_color="lightgray"
        )
    )
    
    gt_net.show()

Forest Service Monthly Net Change (2016-2025)
Month	2016	2017	2018	2019	2020	2021	2022	2023	2024	2025
Jan	−167	251	−211	−299	−182	−153	−89	290		−255
Feb	29	−65	−17	−134	−33	64	90	99		−168
Mar	433	108	86	520	1,078	561	684	724		68
Apr	1,708	4,094	3,694	3,366	2,338	2,623	2,570	2,522	1,510
May	7,366	4,879	5,001	4,618	5,266	4,190	3,206	2,937	2,538
Jun	970	1,011	888	1,074	1,140	787	1,068	1,671	1,392
Jul	103	−20	53	−33	−38	−314	390	407	−148
Aug	−1,351	−1,315	−1,195	−1,208	−856	−1,102	−800	−840	−892
Sep	−1,955	−2,549	−2,177	−2,027	−1,389	−1,474	−1,254	−865	−1,004
Oct	−4,103	−3,685	−3,398	−2,935	−2,432	−2,654	−1,754		−1,642
Nov	−2,128	−2,342	−2,238	−2,163	−2,306	−2,017	−1,479		−1,137
Dec	−716	−675	−697	−852	−873	−963	−757		−480

Annual Trends (Complete Years Only)

Show code

if not monthly_combined.empty:
    # Calculate annual totals and months per year
    annual_summary = monthly_combined.groupby('year').agg({
        'accessions': 'sum',
        'separations': 'sum',
        'net_change': 'sum',
        'month': 'count'  # Count of months with data
    }).reset_index()
    
    # Only include years with 12 months of data
    complete_years = annual_summary[annual_summary['month'] == 12].copy()
    
    if not complete_years.empty:
        print(f"Showing {len(complete_years)} complete years with 12 months of data each")
        print(f"Complete years: {sorted(complete_years['year'].tolist())}")
        
        # Create line plot
        fig, ax = plt.subplots(figsize=(12, 6))
        
        ax.plot(complete_years['year'], complete_years['accessions'], 
                color=ACCESSIONS_COLOR, linewidth=3, marker='o', markersize=8, label='Accessions')
        ax.plot(complete_years['year'], complete_years['separations'], 
                color=SEPARATIONS_COLOR, linewidth=3, marker='s', markersize=8, label='Separations')
        ax.plot(complete_years['year'], complete_years['net_change'], 
                color='black', linewidth=2, marker='D', markersize=6, label='Net Change', linestyle='--')
        
        # Add zero line
        ax.axhline(y=0, color='gray', linestyle='-', alpha=0.5)
        
        # Formatting
        ax.set_xlabel('Year', fontsize=12)
        ax.set_ylabel('Count', fontsize=12)
        ax.set_title('Forest Service Annual Accessions, Separations, and Net Change (Complete Years Only)', fontsize=14, fontweight='bold')
        ax.grid(True, alpha=0.3)
        ax.legend()
        
        # Format y-axis with commas and ensure integer x-axis
        ax.yaxis.set_major_formatter(plt.FuncFormatter(lambda x, p: f'{int(x):,}'))
        ax.xaxis.set_major_formatter(plt.FuncFormatter(lambda x, p: f'{int(x)}'))
        
        plt.tight_layout()
        plt.show()
    else:
        print("No complete years (12 months of data) found for annual analysis")
        
        # Show what we have
        print("\nYears with partial data:")
        for _, row in annual_summary.iterrows():
            print(f"  {row['year']}: {row['month']} months of data")

Showing 18 complete years with 12 months of data each
Complete years: [2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022]

Complete Historical View (All Years)

Show code

if not monthly_combined.empty:
    # Full historical pivot tables
    accessions_pivot_full = monthly_combined.pivot(index='month', columns='year', values='accessions')
    separations_pivot_full = monthly_combined.pivot(index='month', columns='year', values='separations')
    net_pivot_full = monthly_combined.pivot(index='month', columns='year', values='net_change')
    
    # Apply month labels to all
    for pivot in [accessions_pivot_full, separations_pivot_full, net_pivot_full]:
        pivot.index = [month_labels[i-1] for i in pivot.index]
    
    # Create tiny accessions table
    acc_df_full = accessions_pivot_full.reset_index()
    acc_df_full.columns = ['Month'] + [str(col) for col in accessions_pivot_full.columns]
    
    gt_acc_full = (
        GT(acc_df_full)
        .tab_header(title="Forest Service Monthly Accessions (All Years)")
        .fmt_number(columns=list(acc_df_full.columns[1:]), decimals=0, use_seps=True)
        .data_color(
            columns=list(acc_df_full.columns[1:]),
            palette=["white", ACCESSIONS_COLOR],
            domain=[0, accessions_pivot_full.max().max()],
            na_color="lightgray"
        )
        .tab_options(
            table_font_size="6px",
            data_row_padding="1px",
            column_labels_font_size="5px"
        )
    )
    gt_acc_full.show()
    
    # Create tiny separations table
    sep_df_full = separations_pivot_full.reset_index()
    sep_df_full.columns = ['Month'] + [str(col) for col in separations_pivot_full.columns]
    
    gt_sep_full = (
        GT(sep_df_full)
        .tab_header(title="Forest Service Monthly Separations (All Years)")
        .fmt_number(columns=list(sep_df_full.columns[1:]), decimals=0, use_seps=True)
        .data_color(
            columns=list(sep_df_full.columns[1:]),
            palette=["white", SEPARATIONS_COLOR],
            domain=[0, separations_pivot_full.max().max()],
            na_color="lightgray"
        )
        .tab_options(
            table_font_size="6px",
            data_row_padding="1px",
            column_labels_font_size="5px"
        )
    )
    gt_sep_full.show()
    
    # Create tiny net change table
    net_df_full = net_pivot_full.reset_index()
    net_df_full.columns = ['Month'] + [str(col) for col in net_pivot_full.columns]
    
    max_abs_val_full = max(abs(net_pivot_full.min().min()), abs(net_pivot_full.max().max()))
    
    gt_net_full = (
        GT(net_df_full)
        .tab_header(title="Forest Service Monthly Net Change (All Years)")
        .fmt_number(columns=list(net_df_full.columns[1:]), decimals=0, use_seps=True)
        .data_color(
            columns=list(net_df_full.columns[1:]),
            palette=[NET_NEGATIVE_COLOR, "white", NET_POSITIVE_COLOR],
            domain=[-max_abs_val_full, max_abs_val_full],
            na_color="lightgray"
        )
        .tab_options(
            table_font_size="6px",
            data_row_padding="1px",
            column_labels_font_size="5px"
        )
    )
    gt_net_full.show()

Forest Service Monthly Accessions (All Years)
Month	2004	2005	2006	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	2021	2022	2023	2024	2025
Jan		411	392	266	217	323	909	525	246	86	220	393	350	784	238	83	272	351	285	631		123
Feb		348	482	284	129	276	594	452	206	142	188	257	300	193	247	233	293	309	441	436		111
Mar		589	616	429	914	1,186	1,002	712	388	439	610	805	738	422	556	944	1,377	911	1,062	1,125		502
Apr		1,545	3,046	3,858	2,563	2,748	2,672	1,863	1,859	1,720	1,757	2,008	2,163	4,480	4,088	3,720	2,644	3,035	2,984	2,884	1,949
May		7,855	6,066	4,431	6,349	7,043	7,014	6,070	6,104	5,400	6,209	7,557	7,735	5,245	5,373	4,987	5,556	4,566	3,605	3,225	2,910
Jun		2,772	2,679	2,014	2,988	3,185	3,545	3,140	2,954	2,926	3,003	1,292	1,336	1,354	1,252	1,406	1,450	1,144	1,408	2,046	1,744
Jul		482	497	368	551	635	568	645	565	348	286	341	488	371	398	341	402	288	899	788	221
Aug		353	297	223	440	489	506	186	188	174	206	265	271	235	317	331	503	260	403	415	151
Sep		215	290	273	213	252	215	115	144	139	145	216	198	99	269	251	234	229	304	383	82
Oct	330	317	262	230	186	226	260	103	263	119	165	196	300	151	156	234	215	213	393		102
Nov	204	160	168	163	189	233	241	119	119	83	231	248	139	107	156	178	143	154	267		82
Dec	170	140	129	79	152	224	178	70	84	164	125	147	186	97	88	153	143	116	246		54

Forest Service Monthly Separations (All Years)
Month	2004	2005	2006	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	2021	2022	2023	2024	2025
Jan		635	638	656	679	746	750	495	232	552	613	564	517	533	449	382	454	504	374	341		378
Feb		352	298	273	317	263	293	255	205	248	230	221	271	258	264	367	326	245	351	337		279
Mar		459	414	447	403	271	297	261	265	257	255	309	305	314	470	424	299	350	378	401		434
Apr		604	463	456	339	272	316	309	283	256	316	380	455	386	394	354	306	412	414	362	439
May		587	513	507	412	306	304	337	348	365	459	550	369	366	372	369	290	376	399	288	372
Jun		574	531	535	380	277	360	324	400	371	305	317	366	343	364	332	310	357	340	375	352
Jul		635	509	461	435	492	596	558	371	354	337	399	385	391	345	374	440	602	509	381	369
Aug		3,186	3,276	2,873	3,657	3,523	3,395	2,539	2,472	1,566	1,645	1,509	1,622	1,550	1,512	1,539	1,359	1,362	1,203	1,255	1,043
Sep		3,127	3,071	3,070	2,654	2,641	3,074	2,826	2,181	2,224	1,939	1,809	2,153	2,648	2,446	2,278	1,623	1,703	1,558	1,248	1,086
Oct	4,436	3,787	2,719	2,507	2,578	3,672	3,929	3,702	3,161	3,192	3,505	4,291	4,403	3,836	3,554	3,169	2,647	2,867	2,147		1,744
Nov	1,801	2,064	2,006	2,057	2,308	2,575	2,480	2,061	2,796	2,666	3,185	2,527	2,267	2,449	2,394	2,341	2,449	2,171	1,746		1,219
Dec	1,143	798	741	797	795	944	1,102	1,122	1,130	624	709	755	902	772	785	1,005	1,016	1,079	1,003		534

Forest Service Monthly Net Change (All Years)
Month	2004	2005	2006	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	2021	2022	2023	2024	2025
Jan		−224	−246	−390	−462	−423	159	30	14	−466	−393	−171	−167	251	−211	−299	−182	−153	−89	290		−255
Feb		−4	184	11	−188	13	301	197	1	−106	−42	36	29	−65	−17	−134	−33	64	90	99		−168
Mar		130	202	−18	511	915	705	451	123	182	355	496	433	108	86	520	1,078	561	684	724		68
Apr		941	2,583	3,402	2,224	2,476	2,356	1,554	1,576	1,464	1,441	1,628	1,708	4,094	3,694	3,366	2,338	2,623	2,570	2,522	1,510
May		7,268	5,553	3,924	5,937	6,737	6,710	5,733	5,756	5,035	5,750	7,007	7,366	4,879	5,001	4,618	5,266	4,190	3,206	2,937	2,538
Jun		2,198	2,148	1,479	2,608	2,908	3,185	2,816	2,554	2,555	2,698	975	970	1,011	888	1,074	1,140	787	1,068	1,671	1,392
Jul		−153	−12	−93	116	143	−28	87	194	−6	−51	−58	103	−20	53	−33	−38	−314	390	407	−148
Aug		−2,833	−2,979	−2,650	−3,217	−3,034	−2,889	−2,353	−2,284	−1,392	−1,439	−1,244	−1,351	−1,315	−1,195	−1,208	−856	−1,102	−800	−840	−892
Sep		−2,912	−2,781	−2,797	−2,441	−2,389	−2,859	−2,711	−2,037	−2,085	−1,794	−1,593	−1,955	−2,549	−2,177	−2,027	−1,389	−1,474	−1,254	−865	−1,004
Oct	−4,106	−3,470	−2,457	−2,277	−2,392	−3,446	−3,669	−3,599	−2,898	−3,073	−3,340	−4,095	−4,103	−3,685	−3,398	−2,935	−2,432	−2,654	−1,754		−1,642
Nov	−1,597	−1,904	−1,838	−1,894	−2,119	−2,342	−2,239	−1,942	−2,677	−2,583	−2,954	−2,279	−2,128	−2,342	−2,238	−2,163	−2,306	−2,017	−1,479		−1,137
Dec	−973	−658	−612	−718	−643	−720	−924	−1,052	−1,046	−460	−584	−608	−716	−675	−697	−852	−873	−963	−757		−480

Key Insights

Show code

if not monthly_combined.empty:
    # Find months with highest accessions/separations
    max_acc_idx = monthly_combined['accessions'].idxmax()
    max_sep_idx = monthly_combined['separations'].idxmax()
    max_net_pos_idx = monthly_combined[monthly_combined['net_change'] > 0]['net_change'].idxmax() if (monthly_combined['net_change'] > 0).any() else None
    max_net_neg_idx = monthly_combined[monthly_combined['net_change'] < 0]['net_change'].idxmin() if (monthly_combined['net_change'] < 0).any() else None
    
    insights = []
    
    if not pd.isna(max_acc_idx):
        max_acc = monthly_combined.loc[max_acc_idx]
        insights.append({
            'Metric': 'Highest Monthly Accessions',
            'Month': f"{month_labels[int(max_acc['month'])-1]} {int(max_acc['year'])}",
            'Value': f"{int(max_acc['accessions']):,}"
        })
    
    if not pd.isna(max_sep_idx):
        max_sep = monthly_combined.loc[max_sep_idx]
        insights.append({
            'Metric': 'Highest Monthly Separations',
            'Month': f"{month_labels[int(max_sep['month'])-1]} {int(max_sep['year'])}",
            'Value': f"{int(max_sep['separations']):,}"
        })
    
    if max_net_pos_idx is not None and not pd.isna(max_net_pos_idx):
        max_net_pos = monthly_combined.loc[max_net_pos_idx]
        insights.append({
            'Metric': 'Largest Net Gain',
            'Month': f"{month_labels[int(max_net_pos['month'])-1]} {int(max_net_pos['year'])}",
            'Value': f"+{int(max_net_pos['net_change']):,}"
        })
    
    if max_net_neg_idx is not None and not pd.isna(max_net_neg_idx):
        max_net_neg = monthly_combined.loc[max_net_neg_idx]
        insights.append({
            'Metric': 'Largest Net Loss',
            'Month': f"{month_labels[int(max_net_neg['month'])-1]} {int(max_net_neg['year'])}",
            'Value': f"{int(max_net_neg['net_change']):,}"
        })
    
    if insights:
        insights_df = pd.DataFrame(insights)
        
        gt_insights = (
            GT(insights_df)
            .tab_header(title="Key Insights")
        )
        
        gt_insights.show()

Key Insights
Metric	Month	Value
Highest Monthly Accessions	May 2005	7,855
Highest Monthly Separations	Oct 2004	4,436
Largest Net Gain	May 2016	+7,366
Largest Net Loss	Oct 2004	-4,106