기술통계

import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
from scipy import stats
 
# Load Data
orders = pd.read_csv('src_orders.csv', parse_dates=['created_at'])
items = pd.read_csv('src_order_items.csv')
products = pd.read_csv('src_products.csv')
users = pd.read_csv('src_users.csv')
 
# Merge for Analysis
df = orders.merge(items, on='order_id').merge(products, on='product_id').merge(users, on='user_id')

import pandas as pd
import numpy as np
 
# 기본 통계
print("=== 중심 경향 ===")
print(f"평균: ${df['sale_price'].mean():.2f}")
print(f"중앙값: ${df['sale_price'].median():.2f}")
print(f"최빈값: ${df['sale_price'].mode()[0]:.2f}")
 
# describe()로 한번에
print("\n=== describe() ===")
print(df['sale_price'].describe())

SELECT
    AVG(sale_price) as mean_price,
    PERCENTILE_CONT(sale_price, 0.5) OVER() as median_price,
    MIN(sale_price) as min_price,
    MAX(sale_price) as max_price
FROM src_order_items

print("=== 산포도 ===")
print(f"범위: {df['sale_price'].max() - df['sale_price'].min():.2f}")
print(f"분산: {df['sale_price'].var():.2f}")
print(f"표준편차: {df['sale_price'].std():.2f}")
print(f"IQR: {df['sale_price'].quantile(0.75) - df['sale_price'].quantile(0.25):.2f}")
print(f"변동계수: {df['sale_price'].std() / df['sale_price'].mean() * 100:.1f}%")

# 단위가 다른 변수 비교
cv_price = df['sale_price'].std() / df['sale_price'].mean() * 100
cv_age = df['age'].std() / df['age'].mean() * 100
 
print(f"가격 변동계수: {cv_price:.1f}%")
print(f"나이 변동계수: {cv_age:.1f}%")
# 변동계수가 높을수록 상대적으로 더 퍼져있음

from scipy import stats
 
skewness = stats.skew(df['sale_price'].dropna())
print(f"왜도: {skewness:.3f}")
 
if skewness > 0.5:
    print("→ 오른쪽으로 치우침 (고가 제품이 일부 있음)")
elif skewness < -0.5:
    print("→ 왼쪽으로 치우침")
else:
    print("→ 대칭에 가까움")

kurtosis = stats.kurtosis(df['sale_price'].dropna())
print(f"첨도: {kurtosis:.3f}")

# 부서별 기술통계
dept_stats = df.groupby('department')['sale_price'].agg([
    ('개수', 'count'),
    ('평균', 'mean'),
    ('중앙값', 'median'),
    ('표준편차', 'std'),
    ('최소', 'min'),
    ('최대', 'max')
]).round(2)
 
print("부서별 가격 통계:")
print(dept_stats)

SELECT
    department,
    COUNT(*) as count,
    AVG(sale_price) as mean,
    STDDEV(sale_price) as std,
    MIN(sale_price) as min,
    MAX(sale_price) as max
FROM src_order_items oi
JOIN src_products p ON oi.product_id = p.product_id
GROUP BY department
ORDER BY mean DESC