Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,149 +1,359 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
-
import pandas as pd
|
| 3 |
import numpy as np
|
|
|
|
| 4 |
from sklearn.cluster import KMeans
|
| 5 |
-
from sklearn.metrics import
|
| 6 |
import matplotlib.pyplot as plt
|
|
|
|
| 7 |
import io
|
|
|
|
|
|
|
| 8 |
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
# Use policy attributes for clustering
|
| 20 |
-
# Ensure these column names match your policy data CSV
|
| 21 |
-
required_cols = ['IssueAge', 'PolicyTerm', 'SumAssured', 'Duration']
|
| 22 |
-
if not all(col in policy_df.columns for col in required_cols):
|
| 23 |
-
missing_cols = [col for col in required_cols if col not in policy_df.columns]
|
| 24 |
-
return (None, None, None, f"Policy data missing required columns: {missing_cols}. Please ensure your policy CSV has these columns.")
|
| 25 |
-
|
| 26 |
-
X = policy_df[required_cols].fillna(0) # Simple imputation
|
| 27 |
-
|
| 28 |
-
# Handle cases with zero standard deviation (e.g., if a column has all same values after fillna)
|
| 29 |
-
X_std = X.std()
|
| 30 |
-
if (X_std == 0).any():
|
| 31 |
-
zero_std_cols = X_std[X_std == 0].index.tolist()
|
| 32 |
-
return (None, None, None, f"Error: Columns {zero_std_cols} have zero standard deviation after fillna(0). Cannot scale these columns. Please check your data.")
|
| 33 |
|
| 34 |
-
|
| 35 |
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
kmeans.
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
return (
|
| 43 |
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
|
|
|
| 48 |
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
return (None, None, None, "Error: Model point indices not found in PV data. Ensure Policy IDs match.")
|
| 58 |
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
|
|
|
| 63 |
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
|
|
|
| 71 |
|
| 72 |
-
#
|
| 73 |
-
fig, ax = plt.subplots(figsize=(8,4))
|
| 74 |
-
seriatim_cashflows.plot(ax=ax, label='Seriatim Cashflows')
|
| 75 |
-
proxy_cashflows.plot(ax=ax, label='Proxy Cashflows', linestyle='--')
|
| 76 |
-
ax.set_title('Aggregated Cashflows Comparison')
|
| 77 |
-
ax.legend()
|
| 78 |
-
ax.grid(True)
|
| 79 |
-
plt.tight_layout()
|
| 80 |
buf = io.BytesIO()
|
| 81 |
-
plt.savefig(buf, format='png')
|
| 82 |
-
plt.close(fig)
|
| 83 |
buf.seek(0)
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
proxy_pv_df = pv_df.loc[model_points.index]
|
| 88 |
-
# Assuming pv_df has one column of PVs, or sum all columns if multiple
|
| 89 |
-
if proxy_pv_df.shape[1] > 1:
|
| 90 |
-
proxy_pv = proxy_pv_df.multiply(model_points['Weight'].values, axis=0).sum().sum()
|
| 91 |
-
seriatim_pv = pv_df.sum().sum()
|
| 92 |
-
else:
|
| 93 |
-
proxy_pv = proxy_pv_df.multiply(model_points['Weight'].values, axis=0).sum().iloc[0]
|
| 94 |
-
seriatim_pv = pv_df.sum().iloc[0]
|
| 95 |
-
|
| 96 |
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
plt.tight_layout()
|
| 103 |
-
|
| 104 |
-
plt.savefig(
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
# Accuracy metrics
|
| 110 |
-
common_idx = seriatim_cashflows.index.intersection(proxy_cashflows.index)
|
| 111 |
-
if not common_idx.empty:
|
| 112 |
-
r2 = r2_score(seriatim_cashflows.loc[common_idx], proxy_cashflows.loc[common_idx])
|
| 113 |
-
else:
|
| 114 |
-
r2 = float('nan') # Or handle as error
|
| 115 |
-
|
| 116 |
-
pv_error = abs(proxy_pv - seriatim_pv) / seriatim_pv * 100 if seriatim_pv != 0 else float('inf')
|
| 117 |
-
|
| 118 |
-
metrics_text = (
|
| 119 |
-
f"R-squared for aggregated cashflows: {r2:.4f}\n"
|
| 120 |
-
f"Absolute percentage error in present value: {pv_error:.4f}%"
|
| 121 |
-
)
|
| 122 |
-
|
| 123 |
-
return csv_data, cashflow_plot, pv_plot, metrics_text
|
| 124 |
-
|
| 125 |
-
with gr.Blocks() as demo:
|
| 126 |
-
gr.Markdown("# Actuarial Model Point Selection (CSV Upload)")
|
| 127 |
|
| 128 |
-
|
| 129 |
-
|
| 130 |
-
|
| 131 |
-
|
| 132 |
-
|
| 133 |
-
|
| 134 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 135 |
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 141 |
|
| 142 |
-
|
| 143 |
-
|
| 144 |
-
|
| 145 |
-
|
| 146 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 147 |
|
| 148 |
-
if __name__ ==
|
| 149 |
-
demo
|
|
|
|
|
|
| 1 |
import gradio as gr
|
|
|
|
| 2 |
import numpy as np
|
| 3 |
+
import pandas as pd
|
| 4 |
from sklearn.cluster import KMeans
|
| 5 |
+
from sklearn.metrics import pairwise_distances_argmin_min, r2_score
|
| 6 |
import matplotlib.pyplot as plt
|
| 7 |
+
import matplotlib.cm
|
| 8 |
import io
|
| 9 |
+
import base64
|
| 10 |
+
from PIL import Image
|
| 11 |
|
| 12 |
+
class Clusters:
|
| 13 |
+
def __init__(self, loc_vars):
|
| 14 |
+
self.kmeans = kmeans = KMeans(n_clusters=1000, random_state=0, n_init=10).fit(np.ascontiguousarray(loc_vars))
|
| 15 |
+
closest, _ = pairwise_distances_argmin_min(kmeans.cluster_centers_, np.ascontiguousarray(loc_vars))
|
| 16 |
+
|
| 17 |
+
rep_ids = pd.Series(data=(closest+1)) # 0-based to 1-based indexes
|
| 18 |
+
rep_ids.name = 'policy_id'
|
| 19 |
+
rep_ids.index.name = 'cluster_id'
|
| 20 |
+
self.rep_ids = rep_ids
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
+
self.policy_count = self.agg_by_cluster(pd.DataFrame({'policy_count': [1] * len(loc_vars)}))['policy_count']
|
| 23 |
|
| 24 |
+
def agg_by_cluster(self, df, agg=None):
|
| 25 |
+
"""Aggregate columns by cluster"""
|
| 26 |
+
temp = df.copy()
|
| 27 |
+
temp['cluster_id'] = self.kmeans.labels_
|
| 28 |
+
temp = temp.set_index('cluster_id')
|
| 29 |
+
agg = {c: (agg[c] if c in agg else 'sum') for c in temp.columns} if agg else "sum"
|
| 30 |
+
return temp.groupby(temp.index).agg(agg)
|
| 31 |
|
| 32 |
+
def extract_reps(self, df):
|
| 33 |
+
"""Extract the rows of representative policies"""
|
| 34 |
+
temp = pd.merge(self.rep_ids, df.reset_index(), how='left', on='policy_id')
|
| 35 |
+
temp.index.name = 'cluster_id'
|
| 36 |
+
return temp.drop('policy_id', axis=1)
|
| 37 |
|
| 38 |
+
def extract_and_scale_reps(self, df, agg=None):
|
| 39 |
+
"""Extract and scale the rows of representative policies"""
|
| 40 |
+
if agg:
|
| 41 |
+
cols = df.columns
|
| 42 |
+
mult = pd.DataFrame({c: (self.policy_count if (c not in agg or agg[c] == 'sum') else 1) for c in cols})
|
| 43 |
+
return self.extract_reps(df).mul(mult)
|
| 44 |
+
else:
|
| 45 |
+
return self.extract_reps(df).mul(self.policy_count, axis=0)
|
|
|
|
| 46 |
|
| 47 |
+
def compare(self, df, agg=None):
|
| 48 |
+
"""Returns a multi-indexed Dataframe comparing actual and estimate"""
|
| 49 |
+
source = self.agg_by_cluster(df, agg)
|
| 50 |
+
target = self.extract_and_scale_reps(df, agg)
|
| 51 |
+
return pd.DataFrame({'actual': source.stack(), 'estimate':target.stack()})
|
| 52 |
|
| 53 |
+
def compare_total(self, df, agg=None):
|
| 54 |
+
"""Aggregate df by columns"""
|
| 55 |
+
if agg:
|
| 56 |
+
cols = df.columns
|
| 57 |
+
op = {c: (agg[c] if c in agg else 'sum') for c in df.columns}
|
| 58 |
+
actual = df.agg(op)
|
| 59 |
+
estimate = self.extract_and_scale_reps(df, agg=op)
|
| 60 |
+
|
| 61 |
+
op = {k: ((lambda s: s.dot(self.policy_count) / self.policy_count.sum()) if v == 'mean' else v) for k, v in op.items()}
|
| 62 |
+
estimate = estimate.agg(op)
|
| 63 |
+
else:
|
| 64 |
+
actual = df.sum()
|
| 65 |
+
estimate = self.extract_and_scale_reps(df).sum()
|
| 66 |
+
|
| 67 |
+
return pd.DataFrame({'actual': actual, 'estimate': estimate, 'error': estimate / actual - 1})
|
| 68 |
|
| 69 |
+
def create_plot(plot_func, *args, **kwargs):
|
| 70 |
+
"""Helper function to create plots and return as image"""
|
| 71 |
+
plt.figure(figsize=(10, 6))
|
| 72 |
+
plot_func(*args, **kwargs)
|
| 73 |
|
| 74 |
+
# Save plot to bytes
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 75 |
buf = io.BytesIO()
|
| 76 |
+
plt.savefig(buf, format='png', dpi=150, bbox_inches='tight')
|
|
|
|
| 77 |
buf.seek(0)
|
| 78 |
+
plt.close()
|
| 79 |
+
|
| 80 |
+
return Image.open(buf)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 81 |
|
| 82 |
+
def plot_cashflows_comparison(cfs_list, cluster_obj, titles):
|
| 83 |
+
"""Create cashflow comparison plots"""
|
| 84 |
+
fig, axes = plt.subplots(2, 2, figsize=(15, 10))
|
| 85 |
+
axes = axes.flatten()
|
| 86 |
+
|
| 87 |
+
for i, (df, title) in enumerate(zip(cfs_list, titles)):
|
| 88 |
+
if i < len(axes):
|
| 89 |
+
comparison = cluster_obj.compare_total(df)
|
| 90 |
+
comparison[['actual', 'estimate']].plot(ax=axes[i], grid=True, title=title)
|
| 91 |
+
|
| 92 |
plt.tight_layout()
|
| 93 |
+
buf = io.BytesIO()
|
| 94 |
+
plt.savefig(buf, format='png', dpi=150, bbox_inches='tight')
|
| 95 |
+
buf.seek(0)
|
| 96 |
+
plt.close()
|
| 97 |
+
|
| 98 |
+
return Image.open(buf)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 99 |
|
| 100 |
+
def plot_scatter_comparison(df, title):
|
| 101 |
+
"""Create scatter plot comparison"""
|
| 102 |
+
plt.figure(figsize=(12, 8))
|
| 103 |
+
|
| 104 |
+
colors = matplotlib.cm.rainbow(np.linspace(0, 1, len(df.index.levels[1])))
|
| 105 |
+
|
| 106 |
+
for y, c in zip(df.index.levels[1], colors):
|
| 107 |
+
plt.scatter(df.xs(y, level=1)['actual'], df.xs(y, level=1)['estimate'],
|
| 108 |
+
color=c, s=9, alpha=0.6)
|
| 109 |
+
|
| 110 |
+
plt.xlabel('Actual')
|
| 111 |
+
plt.ylabel('Estimate')
|
| 112 |
+
plt.title(title)
|
| 113 |
+
plt.grid(True)
|
| 114 |
+
|
| 115 |
+
# Draw identity line
|
| 116 |
+
lims = [
|
| 117 |
+
np.min([plt.xlim(), plt.ylim()]),
|
| 118 |
+
np.max([plt.xlim(), plt.ylim()]),
|
| 119 |
+
]
|
| 120 |
+
plt.plot(lims, lims, 'r-', linewidth=0.5)
|
| 121 |
+
plt.xlim(lims)
|
| 122 |
+
plt.ylim(lims)
|
| 123 |
+
|
| 124 |
+
buf = io.BytesIO()
|
| 125 |
+
plt.savefig(buf, format='png', dpi=150, bbox_inches='tight')
|
| 126 |
+
buf.seek(0)
|
| 127 |
+
plt.close()
|
| 128 |
+
|
| 129 |
+
return Image.open(buf)
|
| 130 |
|
| 131 |
+
def process_files(cashflow_base, cashflow_lapse, cashflow_mort, policy_data, pv_base, pv_lapse, pv_mort):
|
| 132 |
+
"""Main processing function"""
|
| 133 |
+
try:
|
| 134 |
+
# Read uploaded files
|
| 135 |
+
cfs = pd.read_excel(cashflow_base.name, index_col=0)
|
| 136 |
+
cfs_lapse50 = pd.read_excel(cashflow_lapse.name, index_col=0)
|
| 137 |
+
cfs_mort15 = pd.read_excel(cashflow_mort.name, index_col=0)
|
| 138 |
+
|
| 139 |
+
pol_data = pd.read_excel(policy_data.name, index_col=0)
|
| 140 |
+
if pol_data.shape[1] > 4:
|
| 141 |
+
pol_data = pol_data[['age_at_entry', 'policy_term', 'sum_assured', 'duration_mth']]
|
| 142 |
+
|
| 143 |
+
pvs = pd.read_excel(pv_base.name, index_col=0)
|
| 144 |
+
pvs_lapse50 = pd.read_excel(pv_lapse.name, index_col=0)
|
| 145 |
+
pvs_mort15 = pd.read_excel(pv_mort.name, index_col=0)
|
| 146 |
+
|
| 147 |
+
cfs_list = [cfs, cfs_lapse50, cfs_mort15]
|
| 148 |
+
pvs_list = [pvs, pvs_lapse50, pvs_mort15]
|
| 149 |
+
scen_titles = ['Base', 'Lapse+50%', 'Mort+15%']
|
| 150 |
+
|
| 151 |
+
results = {}
|
| 152 |
+
|
| 153 |
+
# 1. Cashflow Calibration
|
| 154 |
+
cluster_cfs = Clusters(cfs)
|
| 155 |
+
|
| 156 |
+
# Cashflow comparison tables
|
| 157 |
+
results['cf_base_table'] = cluster_cfs.compare_total(cfs)
|
| 158 |
+
results['cf_lapse_table'] = cluster_cfs.compare_total(cfs_lapse50)
|
| 159 |
+
results['cf_mort_table'] = cluster_cfs.compare_total(cfs_mort15)
|
| 160 |
+
|
| 161 |
+
# Policy attributes analysis
|
| 162 |
+
mean_attrs = {'age_at_entry':'mean', 'policy_term':'mean', 'duration_mth':'mean'}
|
| 163 |
+
results['cf_policy_attrs'] = cluster_cfs.compare_total(pol_data, agg=mean_attrs)
|
| 164 |
+
|
| 165 |
+
# Present value analysis
|
| 166 |
+
results['cf_pv_base'] = cluster_cfs.compare_total(pvs)
|
| 167 |
+
results['cf_pv_lapse'] = cluster_cfs.compare_total(pvs_lapse50)
|
| 168 |
+
results['cf_pv_mort'] = cluster_cfs.compare_total(pvs_mort15)
|
| 169 |
+
|
| 170 |
+
# Create plots for cashflow calibration
|
| 171 |
+
results['cf_cashflow_plot'] = plot_cashflows_comparison(cfs_list, cluster_cfs, scen_titles)
|
| 172 |
+
results['cf_scatter_base'] = plot_scatter_comparison(cluster_cfs.compare(cfs), 'Cashflow Calibration - Base Scenario')
|
| 173 |
+
|
| 174 |
+
# 2. Policy Attribute Calibration
|
| 175 |
+
loc_vars = (pol_data - pol_data.min()) / (pol_data.max() - pol_data.min())
|
| 176 |
+
cluster_attrs = Clusters(loc_vars)
|
| 177 |
+
|
| 178 |
+
results['attr_cf_base'] = cluster_attrs.compare_total(cfs)
|
| 179 |
+
results['attr_policy_attrs'] = cluster_attrs.compare_total(pol_data, agg=mean_attrs)
|
| 180 |
+
results['attr_pv_base'] = cluster_attrs.compare_total(pvs)
|
| 181 |
+
results['attr_cashflow_plot'] = plot_cashflows_comparison(cfs_list, cluster_attrs, scen_titles)
|
| 182 |
+
results['attr_scatter_base'] = plot_scatter_comparison(cluster_attrs.compare(cfs), 'Policy Attribute Calibration - Base Scenario')
|
| 183 |
+
|
| 184 |
+
# 3. Present Value Calibration
|
| 185 |
+
cluster_pvs = Clusters(pvs)
|
| 186 |
+
|
| 187 |
+
results['pv_cf_base'] = cluster_pvs.compare_total(cfs)
|
| 188 |
+
results['pv_policy_attrs'] = cluster_pvs.compare_total(pol_data, agg=mean_attrs)
|
| 189 |
+
results['pv_pv_base'] = cluster_pvs.compare_total(pvs)
|
| 190 |
+
results['pv_pv_lapse'] = cluster_pvs.compare_total(pvs_lapse50)
|
| 191 |
+
results['pv_pv_mort'] = cluster_pvs.compare_total(pvs_mort15)
|
| 192 |
+
results['pv_cashflow_plot'] = plot_cashflows_comparison(cfs_list, cluster_pvs, scen_titles)
|
| 193 |
+
results['pv_scatter_base'] = plot_scatter_comparison(cluster_pvs.compare(pvs), 'Present Value Calibration - Base Scenario')
|
| 194 |
+
|
| 195 |
+
# Summary comparison plot
|
| 196 |
+
fig, ax = plt.subplots(figsize=(12, 8))
|
| 197 |
+
comparison_data = {
|
| 198 |
+
'Cashflow Calibration': [
|
| 199 |
+
abs(cluster_cfs.compare_total(cfs)['error'].mean()),
|
| 200 |
+
abs(cluster_cfs.compare_total(pvs)['error'].mean())
|
| 201 |
+
],
|
| 202 |
+
'Policy Attribute Calibration': [
|
| 203 |
+
abs(cluster_attrs.compare_total(cfs)['error'].mean()),
|
| 204 |
+
abs(cluster_attrs.compare_total(pvs)['error'].mean())
|
| 205 |
+
],
|
| 206 |
+
'Present Value Calibration': [
|
| 207 |
+
abs(cluster_pvs.compare_total(cfs)['error'].mean()),
|
| 208 |
+
abs(cluster_pvs.compare_total(pvs)['error'].mean())
|
| 209 |
+
]
|
| 210 |
+
}
|
| 211 |
+
|
| 212 |
+
x = np.arange(2)
|
| 213 |
+
width = 0.25
|
| 214 |
+
|
| 215 |
+
ax.bar(x - width, comparison_data['Cashflow Calibration'], width, label='Cashflow Calibration')
|
| 216 |
+
ax.bar(x, comparison_data['Policy Attribute Calibration'], width, label='Policy Attribute Calibration')
|
| 217 |
+
ax.bar(x + width, comparison_data['Present Value Calibration'], width, label='Present Value Calibration')
|
| 218 |
+
|
| 219 |
+
ax.set_ylabel('Mean Absolute Error')
|
| 220 |
+
ax.set_title('Calibration Method Comparison')
|
| 221 |
+
ax.set_xticks(x)
|
| 222 |
+
ax.set_xticklabels(['Cashflows', 'Present Values'])
|
| 223 |
+
ax.legend()
|
| 224 |
+
ax.grid(True, alpha=0.3)
|
| 225 |
+
|
| 226 |
+
buf = io.BytesIO()
|
| 227 |
+
plt.savefig(buf, format='png', dpi=150, bbox_inches='tight')
|
| 228 |
+
buf.seek(0)
|
| 229 |
+
plt.close()
|
| 230 |
+
results['summary_plot'] = Image.open(buf)
|
| 231 |
+
|
| 232 |
+
return results
|
| 233 |
+
|
| 234 |
+
except Exception as e:
|
| 235 |
+
return {"error": f"Error processing files: {str(e)}"}
|
| 236 |
|
| 237 |
+
def create_interface():
|
| 238 |
+
with gr.Blocks(title="Cluster Model Points Analysis", theme=gr.themes.Soft()) as demo:
|
| 239 |
+
gr.Markdown("""
|
| 240 |
+
# Cluster Model Points Analysis
|
| 241 |
+
|
| 242 |
+
This application applies cluster analysis to model point selection for insurance portfolios.
|
| 243 |
+
Upload your Excel files to analyze cashflows, policy attributes, and present values using different calibration methods.
|
| 244 |
+
|
| 245 |
+
**Required Files:**
|
| 246 |
+
- 3 Cashflow files (Base, Lapse stress, Mortality stress scenarios)
|
| 247 |
+
- 1 Policy data file
|
| 248 |
+
- 3 Present value files (Base, Lapse stress, Mortality stress scenarios)
|
| 249 |
+
""")
|
| 250 |
+
|
| 251 |
+
with gr.Row():
|
| 252 |
+
with gr.Column():
|
| 253 |
+
gr.Markdown("### Upload Files")
|
| 254 |
+
cashflow_base = gr.File(label="Cashflows - Base Scenario", file_types=[".xlsx"])
|
| 255 |
+
cashflow_lapse = gr.File(label="Cashflows - Lapse Stress (+50%)", file_types=[".xlsx"])
|
| 256 |
+
cashflow_mort = gr.File(label="Cashflows - Mortality Stress (+15%)", file_types=[".xlsx"])
|
| 257 |
+
policy_data = gr.File(label="Policy Data", file_types=[".xlsx"])
|
| 258 |
+
pv_base = gr.File(label="Present Values - Base Scenario", file_types=[".xlsx"])
|
| 259 |
+
pv_lapse = gr.File(label="Present Values - Lapse Stress", file_types=[".xlsx"])
|
| 260 |
+
pv_mort = gr.File(label="Present Values - Mortality Stress", file_types=[".xlsx"])
|
| 261 |
+
|
| 262 |
+
analyze_btn = gr.Button("Analyze", variant="primary", size="lg")
|
| 263 |
+
|
| 264 |
+
with gr.Tabs():
|
| 265 |
+
with gr.TabItem("Summary"):
|
| 266 |
+
summary_plot = gr.Image(label="Calibration Methods Comparison")
|
| 267 |
+
|
| 268 |
+
with gr.TabItem("Cashflow Calibration"):
|
| 269 |
+
gr.Markdown("### Results using Annual Cashflows as Calibration Variables")
|
| 270 |
+
|
| 271 |
+
with gr.Row():
|
| 272 |
+
cf_base_table = gr.Dataframe(label="Base Scenario Comparison")
|
| 273 |
+
cf_policy_attrs = gr.Dataframe(label="Policy Attributes Comparison")
|
| 274 |
+
|
| 275 |
+
cf_cashflow_plot = gr.Image(label="Cashflow Comparisons Across Scenarios")
|
| 276 |
+
cf_scatter_base = gr.Image(label="Scatter Plot - Base Scenario")
|
| 277 |
+
|
| 278 |
+
with gr.Row():
|
| 279 |
+
cf_pv_base = gr.Dataframe(label="Present Values - Base")
|
| 280 |
+
cf_pv_lapse = gr.Dataframe(label="Present Values - Lapse Stress")
|
| 281 |
+
cf_pv_mort = gr.Dataframe(label="Present Values - Mortality Stress")
|
| 282 |
+
|
| 283 |
+
with gr.TabItem("Policy Attribute Calibration"):
|
| 284 |
+
gr.Markdown("### Results using Policy Attributes as Calibration Variables")
|
| 285 |
+
|
| 286 |
+
with gr.Row():
|
| 287 |
+
attr_cf_base = gr.Dataframe(label="Cashflows - Base Scenario")
|
| 288 |
+
attr_policy_attrs = gr.Dataframe(label="Policy Attributes Comparison")
|
| 289 |
+
|
| 290 |
+
attr_cashflow_plot = gr.Image(label="Cashflow Comparisons Across Scenarios")
|
| 291 |
+
attr_scatter_base = gr.Image(label="Scatter Plot - Base Scenario")
|
| 292 |
+
attr_pv_base = gr.Dataframe(label="Present Values - Base Scenario")
|
| 293 |
+
|
| 294 |
+
with gr.TabItem("Present Value Calibration"):
|
| 295 |
+
gr.Markdown("### Results using Present Values as Calibration Variables")
|
| 296 |
+
|
| 297 |
+
with gr.Row():
|
| 298 |
+
pv_cf_base = gr.Dataframe(label="Cashflows - Base Scenario")
|
| 299 |
+
pv_policy_attrs = gr.Dataframe(label="Policy Attributes Comparison")
|
| 300 |
+
|
| 301 |
+
pv_cashflow_plot = gr.Image(label="Cashflow Comparisons Across Scenarios")
|
| 302 |
+
pv_scatter_base = gr.Image(label="Scatter Plot - Base Scenario")
|
| 303 |
+
|
| 304 |
+
with gr.Row():
|
| 305 |
+
pv_pv_base = gr.Dataframe(label="Present Values - Base")
|
| 306 |
+
pv_pv_lapse = gr.Dataframe(label="Present Values - Lapse Stress")
|
| 307 |
+
pv_pv_mort = gr.Dataframe(label="Present Values - Mortality Stress")
|
| 308 |
+
|
| 309 |
+
def update_interface(cashflow_base, cashflow_lapse, cashflow_mort, policy_data, pv_base, pv_lapse, pv_mort):
|
| 310 |
+
if not all([cashflow_base, cashflow_lapse, cashflow_mort, policy_data, pv_base, pv_lapse, pv_mort]):
|
| 311 |
+
return [None] * 17
|
| 312 |
+
|
| 313 |
+
results = process_files(cashflow_base, cashflow_lapse, cashflow_mort, policy_data, pv_base, pv_lapse, pv_mort)
|
| 314 |
+
|
| 315 |
+
if "error" in results:
|
| 316 |
+
gr.Warning(results["error"])
|
| 317 |
+
return [None] * 17
|
| 318 |
+
|
| 319 |
+
return [
|
| 320 |
+
results.get('summary_plot'),
|
| 321 |
+
results.get('cf_base_table'),
|
| 322 |
+
results.get('cf_policy_attrs'),
|
| 323 |
+
results.get('cf_cashflow_plot'),
|
| 324 |
+
results.get('cf_scatter_base'),
|
| 325 |
+
results.get('cf_pv_base'),
|
| 326 |
+
results.get('cf_pv_lapse'),
|
| 327 |
+
results.get('cf_pv_mort'),
|
| 328 |
+
results.get('attr_cf_base'),
|
| 329 |
+
results.get('attr_policy_attrs'),
|
| 330 |
+
results.get('attr_cashflow_plot'),
|
| 331 |
+
results.get('attr_scatter_base'),
|
| 332 |
+
results.get('attr_pv_base'),
|
| 333 |
+
results.get('pv_cf_base'),
|
| 334 |
+
results.get('pv_policy_attrs'),
|
| 335 |
+
results.get('pv_cashflow_plot'),
|
| 336 |
+
results.get('pv_scatter_base'),
|
| 337 |
+
results.get('pv_pv_base'),
|
| 338 |
+
results.get('pv_pv_lapse'),
|
| 339 |
+
results.get('pv_pv_mort')
|
| 340 |
+
]
|
| 341 |
+
|
| 342 |
+
analyze_btn.click(
|
| 343 |
+
update_interface,
|
| 344 |
+
inputs=[cashflow_base, cashflow_lapse, cashflow_mort, policy_data, pv_base, pv_lapse, pv_mort],
|
| 345 |
+
outputs=[
|
| 346 |
+
summary_plot,
|
| 347 |
+
cf_base_table, cf_policy_attrs, cf_cashflow_plot, cf_scatter_base,
|
| 348 |
+
cf_pv_base, cf_pv_lapse, cf_pv_mort,
|
| 349 |
+
attr_cf_base, attr_policy_attrs, attr_cashflow_plot, attr_scatter_base, attr_pv_base,
|
| 350 |
+
pv_cf_base, pv_policy_attrs, pv_cashflow_plot, pv_scatter_base,
|
| 351 |
+
pv_pv_base, pv_pv_lapse, pv_pv_mort
|
| 352 |
+
]
|
| 353 |
+
)
|
| 354 |
+
|
| 355 |
+
return demo
|
| 356 |
|
| 357 |
+
if __name__ == "__main__":
|
| 358 |
+
demo = create_interface()
|
| 359 |
+
demo.launch()
|