data-sci-project/buildings_and_incomes_by_postal_code

1import pandas as pd
2import matplotlib.pyplot as plt
3import seaborn as sns
4
5OUT = "eda_outputs"
6import os
7os.makedirs(OUT, exist_ok=True)
8
9df = pd.read_csv("PaavoDataset.csv", encoding="latin1", sep=";")
10df.columns = df.columns.str.strip()  # remove stray leading/trailing spaces
11df = df[~df["Area"].str.strip().str.startswith("MK")].copy() # Drop regional summary rows (e.g. "MK01 Uusimaa")
12
13#Summary stats
14print("--- Info ---")
15print(df.info())
16print("\n--- Descriptive statistics ---")
17desc = df.describe().T
18print(desc)
19desc.to_csv(f"{OUT}/summary_statistics.csv")
20
21# Distributions of key variables
22key_vars = [
23    "Inhabitants",
24    "Average age of inhabitants",
25    "Average income of inhabitants",
26    "Median income of inhabitants",
27    "Average size of households",
28]
29
30fig, axes = plt.subplots(len(key_vars), 1, figsize=(8, 4 * len(key_vars)))
31for ax, col in zip(axes, key_vars):
32    sns.histplot(df[col], kde=True, ax=ax, color="#6fa8dc")
33    ax.set_title(f"Distribution of {col}")
34plt.tight_layout()
35fig.savefig(f"{OUT}/01_distributions.png", dpi=150)
36plt.close(fig)
37
38# Correlation heatmap (numeric columns only)
39numeric_df = df.select_dtypes(include="number")
40corr = numeric_df.corr()
41
42fig, ax = plt.subplots(figsize=(14, 12))
43sns.heatmap(corr, cmap="coolwarm", center=0, annot=False, ax=ax)
44ax.set_title("Correlation heatmap")
45fig.savefig(f"{OUT}/02_correlation_heatmap.png", dpi=150, bbox_inches="tight")
46plt.close(fig)
47
48# Top areas by income and population
49top_income = df.nlargest(15, "Average income of inhabitants")[["Area", "Average income of inhabitants"]]
50top_population = df.nlargest(15, "Inhabitants")[["Area", "Inhabitants"]]
51print("\n--- Top 15 areas by average income ---")
52print(top_income)
53print("\n--- Top 15 areas by population ---")
54print(top_population)
55
56# Income category composition (stacked bar, top 20 by inhabitants)
57income_cats = [
58    "Inhabitants belonging to the lowest income category",
59    "Inhabitants belonging to the middle income category",
60    "Inhabitants belonging to the highest income category",
61]
62
63top20_income = df.nlargest(20, "Inhabitants").set_index("Area")[income_cats].iloc[::-1]
64top20_income.columns = ["Lowest income cat.", "Middle income cat.", "Highest income cat."]
65
66fig, ax = plt.subplots(figsize=(10, 8))
67top20_income.plot(kind="barh", stacked=True, color=["#6fa8dc", "#cc6659", "#e69138"], ax=ax)
68ax.set_title("Income category composition - top 20 areas by population")
69ax.set_xlabel("Number of inhabitants")
70fig.savefig(f"{OUT}/03_income_categories_top20.png", dpi=150, bbox_inches="tight")
71plt.close(fig)
72
73# Household income category composition (same style, top 20 by households)
74household_cats = [
75    "Households belonging to the lowest income category",
76    "Households belonging to the middle income category",
77    "Households belonging to the highest income category",
78]
79
80top20_households = (df.nlargest(20, "Households").set_index("Area")[household_cats].iloc[::-1])
81top20_households.columns = ["Lowest income cat.", "Middle income cat.", "Highest income cat."]
82
83fig, ax = plt.subplots(figsize=(10, 8))
84top20_households.plot(kind="barh", stacked=True, color=["#6fa8dc", "#cc6659", "#e69138"], ax=ax)
85ax.set_title("Household income category composition - top 20 areas by number of households")
86ax.set_xlabel("Number of households")
87fig.savefig(f"{OUT}/04_household_income_categories_top20.png", dpi=150, bbox_inches="tight")
88plt.close(fig)
89 
90# Inhabitants per residential building (housing density proxy)
91df["Inhabitants per residential building"] = (
92    df["Inhabitants"] / df["Residential buildings"].replace(0, pd.NA)
93)
94 
95density_top20 = df.nlargest(20, "Inhabitants per residential building")[
96    ["Area", "Inhabitants per residential building"]
97].set_index("Area")
98 
99fig, ax = plt.subplots(figsize=(9, 8))
100density_top20.sort_values("Inhabitants per residential building").plot(
101    kind="barh", legend=False, color="#6fa8dc", ax=ax
102)
103ax.set_title("Top 20 areas by inhabitants per residential building)")
104ax.set_xlabel("Inhabitants per residential building")
105fig.savefig(f"{OUT}/05_inhabitants_per_building_top20.png", dpi=150, bbox_inches="tight")
106plt.close(fig)
107 
108#  Accumulated purchasing power of inhabitants
109purchasing_top20 = df.nlargest(20, "Accumulated purchasing power of inhabitants")[
110    ["Area", "Accumulated purchasing power of inhabitants"]
111].set_index("Area")
112 
113fig, ax = plt.subplots(figsize=(9, 8))
114purchasing_top20.sort_values("Accumulated purchasing power of inhabitants").plot(
115    kind="barh", legend=False, color="#e69138", ax=ax
116)
117ax.set_title("Top 20 areas by accumulated purchasing power of inhabitants")
118ax.set_xlabel("Accumulated purchasing power (EUR)")
119fig.savefig(f"{OUT}/06_purchasing_power_top20.png", dpi=150, bbox_inches="tight")
120plt.close(fig)
121 
122print(f"\nAll plots and summary CSV saved to ./{OUT}/")