Python分析宝马全球销售数据？实战项目教你掌握数据清洗与可视化

import pandas as pd

# 读取多源数据
annual_report = pd.read_excel('bmw_global_sales.xlsx', sheet_name='Regional')
model_details = pd.read_csv('bmw_models_2015-2021.csv')
gdp_data = pd.read_json('world_gdp.json')

# 查看数据结构
print(annual_report.info())
print(model_details.head(3))

# 统一转换为实际销量
annual_report['Europe'] = annual_report['Europe'] * 1000

model_details['Model'] = model_details['Model'].str.replace(' xDrive', '')

# 使用移动平均法补全
annual_report['Europe'].fillna(
    annual_report['Europe'].rolling(3, min_periods=1).mean(),
    inplace=True
)

q1 = annual_report['China'].quantile(0.25)
q3 = annual_report['China'].quantile(0.75)
iqr = q3 - q1
annual_report = annual_report[
    (annual_report['China'] > (q1 - 1.5*iqr)) & 
    (annual_report['China'] < (q3 + 1.5*iqr))
]

model_details['Date'] = pd.to_datetime(model_details['Date'], 
                                      format='mixed').dt.strftime('%Y-%m')

import matplotlib.pyplot as plt
import seaborn as sns

plt.figure(figsize=(12, 6))
sns.set_style("whitegrid")

# 准备数据
regional = annual_report.melt(id_vars=['Year'], 
                            value_vars=['China', 'USA', 'Europe'],
                            var_name='Region', 
                            value_name='Sales')

# 绘制堆叠图
sns.lineplot(data=regional, x='Year', y='Sales', hue='Region', 
             style='Region', markers=True, dashes=False, 
             palette=['#E63946', '#457B9D', '#1D3557'])

plt.title('BMW Regional Sales Trend (2018-2022)', pad=20)
plt.xlabel('Year', labelpad=10)
plt.ylabel('Sales Volume (units)', labelpad=10)
plt.legend(title='Region', bbox_to_anchor=(1.05, 1))
plt.tight_layout()

import plotly.express as px

# 按车型级别和驱动类型聚合
model_analysis = model_details.groupby(['Class', 'Powertrain'])['Sales'].sum().reset_index()

fig = px.sunburst(model_analysis, path=['Class', 'Powertrain'], values='Sales',
                  color='Class', hover_data=['Powertrain'],
                  color_discrete_sequence=px.colors.qualitative.Pastel)

fig.update_layout(title_text='BMW Sales Structure by Class & Powertrain (2015-2021)',
                 margin=dict(t=30, l=0, r=0, b=0))
fig.show()

from scipy.stats import pearsonr

# 合并经济数据
merged = pd.merge(model_details, gdp_data, left_on='Country', right_on='code')

plt.figure(figsize=(10, 8))
sns.pairplot(merged[['Price', 'Sales', 'gdp_per_capita']], 
             diag_kind='kde', 
             plot_kws={'alpha':0.6})

# 计算并标注相关系数
corr, _ = pearsonr(merged['Price'], merged['gdp_per_capita'])
plt.text(0.5, 0.9, f'Price-GDP Correlation: {corr:.2f}', 
         ha='center', transform=plt.gcf().transFigure)

from prophet import Prophet

# 准备中国区数据
china_data = annual_report[['Year', 'China']].rename(
    columns={'Year': 'ds', 'China': 'y'})

# 建模预测
model = Prophet(seasonality_mode='multiplicative')
model.fit(china_data)
future = model.make_future_dataframe(periods=12, freq='M')
forecast = model.predict(future)

# 可视化
fig = model.plot(forecast, xlabel='Date', ylabel='Sales')
plt.title('China Sales Forecast with Prophet')

from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler

# 选择特征列
features = ['Age', 'Income', 'Education', 'Previous_Owned']
X = model_details[features].dropna()

# 数据标准化
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

# 肘部法则确定最佳K值
inertia = []
for k in range(2, 8):
    kmeans = KMeans(n_clusters=k, random_state=42)
    kmeans.fit(X_scaled)
    inertia.append(kmeans.inertia_)

# 可视化肘部曲线
plt.plot(range(2, 8), inertia, marker='o')
plt.xlabel('Number of clusters')
plt.ylabel('Inertia')

from jinja2 import Environment, FileSystemLoader

env = Environment(loader=FileSystemLoader('templates'))
template = env.get_template('report_template.html')

report_data = {
    'top_models': top_models.to_dict('records'),
    'regional_growth': regional_growth,
    'forecast_summary': forecast[['ds', 'yhat']].tail(12)
}

html_report = template.render(report_data)
with open('bmw_sales_report.html', 'w') as f:
    f.write(html_report)

import dash
from dash import dcc, html
import plotly.graph_objects as go

app = dash.Dash(__name__)

app.layout = html.Div([
    dcc.Dropdown(id='region-selector',
                 options=[{'label': r, 'value': r} 
                          for r in annual_report.columns[1:4]]),
    dcc.Graph(id='sales-trend'),
    dcc.Graph(id='model-mix')
])

@app.callback(
    [Output('sales-trend', 'figure'),
     Output('model-mix', 'figure')],
    [Input('region-selector', 'value')]
)
def update_plots(selected_region):
    trend_fig = px.line(annual_report, x='Year', y=selected_region)
    mix_fig = px.pie(model_details[model_details['Region']==selected_region],
                     names='Model', values='Sales')
    return trend_fig, mix_fig