Python数据分析与可视化的全面指南(从数据清洗到图表呈现)

import pandas as pd
 
# 读取CSV文件时自动推断数据类型
df = pd.read_csv('sales_data.csv', parse_dates=['order_date'])
 
# 基础信息概览
print(df.info())        # 内存占用、非空值统计
print(df.describe())    # 数值列统计指标
print(df.head(3))       # 预览前3行结构

# 删除策略：当某列缺失率超过80%时
df.dropna(thresh=len(df)*0.2, axis=1, inplace=True)
 
# 填充策略：时间序列用前向填充
df['price'].fillna(method='ffill', inplace=True)
 
# 业务填充：用户年龄用中位数，地区用众数
df['age'].fillna(df['age'].median(), inplace=True)
df['region'].fillna(df['region'].mode()[0], inplace=True)

# 3σ原则检测数值异常
z_scores = (df['price'] - df['price'].mean()) / df['price'].std()
df = df[(z_scores > -3) & (z_scores < 3)]
 
# 箱线图可视化辅助判断
import seaborn as sns
sns.boxplot(x=df['price'])

# 字符串转分类变量
df['product_category'] = df['product_category'].astype('category')
 
# 拆分复合字段
df[['province', 'city']] = df['location'].str.split('/', expand=True)
 
# 时间戳特征工程
df['hour'] = df['order_time'].dt.hour
df['weekday'] = df['order_time'].dt.weekday

# 构建销售透 视表
pivot = df.pivot_table(
    index='region', 
    columns='product_category',
    values='amount',
    aggfunc='sum',
    fill_value=0
)
 
# 交叉表分析用户行为
cross = pd.crosstab(
    df['device_type'],
    df['purchase_flag'],
    normalize='index'
)

# 计算移动平均销量
df['rolling_avg'] = df.groupby('product_id')['sales'].transform(
    lambda x: x.rolling(7, min_periods=1).mean()
)
 
# 排名函数应用
df['sales_rank'] = df.groupby('region')['sales'].rank(ascending=False)

# 正则表达式清洗
df['clean_comment'] = df['raw_comment'].str.replace(r'[^a-zA-Z0-9\s]', '')
 
# 词频统计基础
from collections import Counter
word_counts = Counter(' '.join(df['clean_comment']).split())

import matplotlib.pyplot as plt
 
# 双Y轴组合图表
fig, ax1 = plt.subplots(figsize=(10,6))
ax1.plot(df['date'], df['sales'], 'b-', label='销售额')
ax2 = ax1.twinx()
ax2.plot(df['date'], df['conversion_rate'], 'r--', label='转化率')
 
# 动态标注极值点
max_idx = df['sales'].idxmax()
ax1.annotate(f'峰值: {df.loc[max_idx,"sales"]}',
            xy=(max_idx, df.loc[max_idx,"sales"]),
            xytext=(max_idx-3, df.loc[max_idx,"sales"]*0.9),
            arrowprops=dict(arrowstyle='->'))

import plotly.express as px
 
# 交互式热力图
fig = px.density_heatmap(
    df, 
    x='order_hour', 
    y='day_of_week',
    z='order_count',
    title='订单时间分布热力图'
)
fig.show()
 
# 桑基图展示用户流转
fig = px.sankey(
    df,
    node={'label': ['访问','加购','下单','支付'],
          'color': ['lightblue','lightgreen','orange','red']},
    source=[0,0,1,1,2],
    target=[1,2,2,3,3],
    value=[800,200,150,50,100]
)

# 数据读取优化
df = pd.read_parquet('user_behavior.parquet', engine='pyarrow')
 
# 行为路径分析
behavior_flow = df.groupby(['user_id'])['behavior_type'].apply(
    lambda x: '/'.join(x.unique())
).value_counts().sort_index()
 
# 漏斗转化计算
funnel = df[df['behavior_type'].isin(['click','cart','buy'])].groupby(
    'session_id'
)['behavior_type'].agg(lambda x: '/'.join(x)).value_counts()
 
# 绘制漏斗图
from pyecharts.charts import Funnel
funnel = Funnel()
funnel.add("转化路径", 
           [list(z) for z in zip(funnel.index, funnel.values)],
           sort_="descending",
           label_opts={"position": "inside"})
funnel.render("conversion_funnel.html")