piechart.py

import pandas as pd
import matplotlib.pyplot as plt

# ---------- Load Data ----------
df = pd.read_csv("montana_label_with_gpe_sorted.csv", usecols=["label"])
df = df.dropna(subset=["label"])

# ---------- Count Label Frequencies ----------
label_counts = df["label"].value_counts()

# ---------- Keep Top 8 Labels, Group Rest as 'Other' ----------
top_labels = label_counts.head(8)
other_count = label_counts[8:].sum()

# Add "Other" if applicable
if other_count > 0:
    top_labels["Other"] = other_count

# ---------- Create Pie Chart ----------
plt.figure(figsize=(8, 8))
plt.pie(
    top_labels,
    labels=top_labels.index,
    autopct='%1.1f%%',
    startangle=140,
    wedgeprops={'edgecolor': 'black'}
)

plt.title("Distribution of Top Labels (Others Grouped)", fontsize=14)
plt.axis('equal')  # Keeps pie chart circular
plt.tight_layout()

# ---------- Save and Show ----------
plt.savefig("label_distribution_pie.png", dpi=300)
plt.show()
print("✅ Pie chart saved as 'label_distribution_pie.png'")