736 lines
24 KiB
Python
736 lines
24 KiB
Python
|
"""
|
|||
|
|
📊 AI 数据分析师
|
|||
|
|
==================
|
|||
|
|
一个基于 DeepSeek API + Streamlit 的智能数据分析应用
|
|||
|
|
|
|||
|
|
功能特性：
|
|||
|
|
- 📁 上传 CSV 数据文件
|
|||
|
|
- 📈 自动生成可视化图表
|
|||
|
|
- 🤖 AI 智能分析数据特征
|
|||
|
|
- 💬 自然语言问答数据
|
|||
|
|
- 📝 生成分析报告
|
|||
|
|
|
|||
|
|
技术栈：
|
|||
|
|
- Streamlit: Web 界面
|
|||
|
|
- Pandas: 数据处理
|
|||
|
|
- Altair: 数据可视化
|
|||
|
|
- DeepSeek API: AI 分析
|
|||
|
|
"""
|
|||
|
|
|
|||
|
|
import streamlit as st
|
|||
|
|
import pandas as pd
|
|||
|
|
import altair as alt
|
|||
|
|
import json
|
|||
|
|
from openai import OpenAI
|
|||
|
|
import os
|
|||
|
|
from dotenv import load_dotenv
|
|||
|
|
from io import StringIO
|
|||
|
|
|
|||
|
|
# 加载环境变量
|
|||
|
|
load_dotenv()
|
|||
|
|
|
|||
|
|
# ============== 配置 ==============
|
|||
|
|
|
|||
|
|
API_KEY = os.getenv("DEEPSEEK_API_KEY", "")
|
|||
|
|
BASE_URL = "https://api.deepseek.com"
|
|||
|
|
MODEL = "deepseek-chat"
|
|||
|
|
|
|||
|
|
# ============== 页面配置 ==============
|
|||
|
|
|
|||
|
|
st.set_page_config(
|
|||
|
|
page_title="AI 数据分析师",
|
|||
|
|
page_icon="📊",
|
|||
|
|
layout="wide",
|
|||
|
|
initial_sidebar_state="expanded"
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
# ============== 自定义样式 ==============
|
|||
|
|
|
|||
|
|
st.markdown("""
|
|||
|
|
<style>
|
|||
|
|
/* 渐变标题 */
|
|||
|
|
.main-title {
|
|||
|
|
background: linear-gradient(120deg, #667eea 0%, #764ba2 100%);
|
|||
|
|
-webkit-background-clip: text;
|
|||
|
|
-webkit-text-fill-color: transparent;
|
|||
|
|
font-size: 2.5rem;
|
|||
|
|
font-weight: 800;
|
|||
|
|
margin-bottom: 0;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/* 统计卡片 */
|
|||
|
|
.stat-card {
|
|||
|
|
background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
|
|||
|
|
border-radius: 15px;
|
|||
|
|
padding: 1.5rem;
|
|||
|
|
color: white;
|
|||
|
|
text-align: center;
|
|||
|
|
box-shadow: 0 10px 30px rgba(102, 126, 234, 0.3);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
.stat-card h2 {
|
|||
|
|
font-size: 2rem;
|
|||
|
|
margin: 0;
|
|||
|
|
font-weight: 700;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
.stat-card p {
|
|||
|
|
margin: 0.5rem 0 0 0;
|
|||
|
|
opacity: 0.9;
|
|||
|
|
font-size: 0.9rem;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/* 洞察卡片 */
|
|||
|
|
.insight-card {
|
|||
|
|
background: linear-gradient(135deg, #f093fb 0%, #f5576c 100%);
|
|||
|
|
border-radius: 15px;
|
|||
|
|
padding: 1.5rem;
|
|||
|
|
color: white;
|
|||
|
|
margin: 1rem 0;
|
|||
|
|
box-shadow: 0 10px 30px rgba(245, 87, 108, 0.3);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/* 分析报告样式 */
|
|||
|
|
.report-section {
|
|||
|
|
background: rgba(102, 126, 234, 0.1);
|
|||
|
|
border-left: 4px solid #667eea;
|
|||
|
|
padding: 1rem 1.5rem;
|
|||
|
|
margin: 1rem 0;
|
|||
|
|
border-radius: 0 10px 10px 0;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/* 问答区域 */
|
|||
|
|
.qa-section {
|
|||
|
|
background: linear-gradient(135deg, #11998e 0%, #38ef7d 100%);
|
|||
|
|
border-radius: 15px;
|
|||
|
|
padding: 1.5rem;
|
|||
|
|
color: white;
|
|||
|
|
margin: 1rem 0;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/* 上传区域美化 */
|
|||
|
|
.uploadedFile {
|
|||
|
|
border: 2px dashed #667eea !important;
|
|||
|
|
border-radius: 15px !important;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/* Tab 样式 */
|
|||
|
|
.stTabs [data-baseweb="tab-list"] {
|
|||
|
|
gap: 8px;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
.stTabs [data-baseweb="tab"] {
|
|||
|
|
border-radius: 10px;
|
|||
|
|
padding: 10px 20px;
|
|||
|
|
}
|
|||
|
|
</style>
|
|||
|
|
""", unsafe_allow_html=True)
|
|||
|
|
|
|||
|
|
# ============== 辅助函数 ==============
|
|||
|
|
|
|||
|
|
def get_data_summary(df: pd.DataFrame) -> str:
|
|||
|
|
"""生成数据摘要文本供 AI 分析"""
|
|||
|
|
summary = []
|
|||
|
|
|
|||
|
|
# 基本信息
|
|||
|
|
summary.append(f"数据集包含 {len(df)} 行和 {len(df.columns)} 列。")
|
|||
|
|
summary.append(f"\n列名：{', '.join(df.columns.tolist())}")
|
|||
|
|
|
|||
|
|
# 数据类型
|
|||
|
|
summary.append(f"\n\n各列数据类型：")
|
|||
|
|
for col in df.columns:
|
|||
|
|
dtype = str(df[col].dtype)
|
|||
|
|
null_count = df[col].isnull().sum()
|
|||
|
|
summary.append(f"- {col}: {dtype}, 缺失值: {null_count}")
|
|||
|
|
|
|||
|
|
# 数值列统计
|
|||
|
|
numeric_cols = df.select_dtypes(include=['number']).columns.tolist()
|
|||
|
|
if numeric_cols:
|
|||
|
|
summary.append(f"\n\n数值列统计摘要：")
|
|||
|
|
stats = df[numeric_cols].describe().to_string()
|
|||
|
|
summary.append(stats)
|
|||
|
|
|
|||
|
|
# 分类列信息
|
|||
|
|
cat_cols = df.select_dtypes(include=['object', 'category']).columns.tolist()
|
|||
|
|
if cat_cols:
|
|||
|
|
summary.append(f"\n\n分类列信息：")
|
|||
|
|
for col in cat_cols[:5]: # 最多显示5个
|
|||
|
|
unique_count = df[col].nunique()
|
|||
|
|
top_values = df[col].value_counts().head(5).to_dict()
|
|||
|
|
summary.append(f"- {col}: {unique_count} 个唯一值, 前5个: {top_values}")
|
|||
|
|
|
|||
|
|
# 样本数据
|
|||
|
|
summary.append(f"\n\n前3行数据样本：")
|
|||
|
|
summary.append(df.head(3).to_string())
|
|||
|
|
|
|||
|
|
return "\n".join(summary)
|
|||
|
|
|
|||
|
|
|
|||
|
|
def analyze_with_ai(client: OpenAI, data_summary: str, analysis_type: str = "general") -> str:
|
|||
|
|
"""使用 AI 分析数据"""
|
|||
|
|
|
|||
|
|
prompts = {
|
|||
|
|
"general": """你是一个专业的数据分析师。请根据以下数据摘要，提供全面的数据分析洞察。
|
|||
|
|
|
|||
|
|
请从以下几个方面进行分析：
|
|||
|
|
1. 📊 数据概况：数据的整体情况和质量评估
|
|||
|
|
2. 🔍 关键发现：数据中的重要模式和特征
|
|||
|
|
3. 📈 建议可视化：推荐适合这个数据集的图表类型
|
|||
|
|
4. ⚠️ 注意事项：数据中可能存在的问题或需要注意的地方
|
|||
|
|
5. 💡 进一步分析建议：可以深入探索的方向
|
|||
|
|
|
|||
|
|
请用清晰的结构和通俗易懂的语言回答。""",
|
|||
|
|
|
|||
|
|
"correlation": """你是一个专业的数据分析师。请分析以下数据中各变量之间可能存在的相关性和关联。
|
|||
|
|
|
|||
|
|
请重点分析：
|
|||
|
|
1. 哪些变量之间可能存在正相关或负相关？
|
|||
|
|
2. 是否存在明显的因果关系假设？
|
|||
|
|
3. 建议做哪些相关性分析？
|
|||
|
|
|
|||
|
|
请用通俗易懂的语言解释。""",
|
|||
|
|
|
|||
|
|
"anomaly": """你是一个专业的数据分析师。请分析以下数据中可能存在的异常值或异常模式。
|
|||
|
|
|
|||
|
|
请关注：
|
|||
|
|
1. 数值是否有明显的异常值？
|
|||
|
|
2. 数据分布是否有异常？
|
|||
|
|
3. 是否存在数据质量问题？
|
|||
|
|
4. 建议如何处理这些异常？
|
|||
|
|
|
|||
|
|
请给出具体的分析和建议。"""
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
system_prompt = prompts.get(analysis_type, prompts["general"])
|
|||
|
|
|
|||
|
|
response = client.chat.completions.create(
|
|||
|
|
model=MODEL,
|
|||
|
|
messages=[
|
|||
|
|
{"role": "system", "content": system_prompt},
|
|||
|
|
{"role": "user", "content": f"数据摘要：\n\n{data_summary}"}
|
|||
|
|
],
|
|||
|
|
max_tokens=2000,
|
|||
|
|
temperature=0.7
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
return response.choices[0].message.content
|
|||
|
|
|
|||
|
|
|
|||
|
|
def ask_data_question(client: OpenAI, data_summary: str, question: str, chat_history: list) -> str:
|
|||
|
|
"""根据数据回答用户问题"""
|
|||
|
|
|
|||
|
|
system_prompt = """你是一个专业的数据分析助手。用户会就数据集提出问题，你需要根据提供的数据摘要来回答。
|
|||
|
|
|
|||
|
|
回答要求：
|
|||
|
|
1. 基于数据事实回答，不要编造数据中没有的信息
|
|||
|
|
2. 如果数据摘要中没有足够信息回答问题，请诚实说明
|
|||
|
|
3. 尽可能提供具体的数字和分析
|
|||
|
|
4. 可以给出进一步分析的建议"""
|
|||
|
|
|
|||
|
|
messages = [{"role": "system", "content": system_prompt}]
|
|||
|
|
messages.append({"role": "user", "content": f"数据摘要：\n{data_summary}\n\n请记住这个数据摘要，我接下来会问你关于这个数据的问题。"})
|
|||
|
|
messages.append({"role": "assistant", "content": "好的，我已经了解了这个数据集的情况。请问您想了解什么？"})
|
|||
|
|
|
|||
|
|
# 添加对话历史
|
|||
|
|
for msg in chat_history:
|
|||
|
|
messages.append(msg)
|
|||
|
|
|
|||
|
|
# 添加当前问题
|
|||
|
|
messages.append({"role": "user", "content": question})
|
|||
|
|
|
|||
|
|
response = client.chat.completions.create(
|
|||
|
|
model=MODEL,
|
|||
|
|
messages=messages,
|
|||
|
|
max_tokens=1500,
|
|||
|
|
temperature=0.7
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
return response.choices[0].message.content
|
|||
|
|
|
|||
|
|
|
|||
|
|
def generate_report(client: OpenAI, data_summary: str, analyses: dict) -> str:
|
|||
|
|
"""生成完整的分析报告"""
|
|||
|
|
|
|||
|
|
system_prompt = """你是一个专业的数据分析师，请根据数据摘要和已有的分析结果，生成一份完整的数据分析报告。
|
|||
|
|
|
|||
|
|
报告格式要求：
|
|||
|
|
1. 使用 Markdown 格式
|
|||
|
|
2. 包含标题、摘要、详细分析、结论和建议
|
|||
|
|
3. 结构清晰，层次分明
|
|||
|
|
4. 语言专业但易懂
|
|||
|
|
5. 适当使用 emoji 增加可读性"""
|
|||
|
|
|
|||
|
|
user_content = f"""数据摘要：
|
|||
|
|
{data_summary}
|
|||
|
|
|
|||
|
|
已有分析结果：
|
|||
|
|
{json.dumps(analyses, ensure_ascii=False, indent=2)}
|
|||
|
|
|
|||
|
|
请生成一份完整的数据分析报告。"""
|
|||
|
|
|
|||
|
|
response = client.chat.completions.create(
|
|||
|
|
model=MODEL,
|
|||
|
|
messages=[
|
|||
|
|
{"role": "system", "content": system_prompt},
|
|||
|
|
{"role": "user", "content": user_content}
|
|||
|
|
],
|
|||
|
|
max_tokens=3000,
|
|||
|
|
temperature=0.7
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
return response.choices[0].message.content
|
|||
|
|
|
|||
|
|
|
|||
|
|
def create_visualizations(df: pd.DataFrame):
|
|||
|
|
"""根据数据自动生成可视化图表"""
|
|||
|
|
charts = []
|
|||
|
|
|
|||
|
|
numeric_cols = df.select_dtypes(include=['number']).columns.tolist()
|
|||
|
|
cat_cols = df.select_dtypes(include=['object', 'category']).columns.tolist()
|
|||
|
|
|
|||
|
|
# 1. 数值分布直方图（第一个数值列）
|
|||
|
|
if numeric_cols:
|
|||
|
|
col = numeric_cols[0]
|
|||
|
|
hist_chart = alt.Chart(df).mark_bar(
|
|||
|
|
opacity=0.7,
|
|||
|
|
color='#667eea'
|
|||
|
|
).encode(
|
|||
|
|
alt.X(f'{col}:Q', bin=alt.Bin(maxbins=30), title=col),
|
|||
|
|
alt.Y('count()', title='频数'),
|
|||
|
|
tooltip=[alt.Tooltip(f'{col}:Q', bin=alt.Bin(maxbins=30)), 'count()']
|
|||
|
|
).properties(
|
|||
|
|
title=f'📊 {col} 分布直方图',
|
|||
|
|
height=300
|
|||
|
|
)
|
|||
|
|
charts.append(('distribution', hist_chart, f'{col} 的数值分布'))
|
|||
|
|
|
|||
|
|
# 2. 分类计数柱状图（第一个分类列）
|
|||
|
|
if cat_cols:
|
|||
|
|
col = cat_cols[0]
|
|||
|
|
if df[col].nunique() <= 20: # 类别不要太多
|
|||
|
|
bar_chart = alt.Chart(df).mark_bar(
|
|||
|
|
color='#764ba2'
|
|||
|
|
).encode(
|
|||
|
|
x=alt.X(f'{col}:N', sort='-y', title=col),
|
|||
|
|
y=alt.Y('count()', title='计数'),
|
|||
|
|
tooltip=[col, 'count()']
|
|||
|
|
).properties(
|
|||
|
|
title=f'📈 {col} 分类计数',
|
|||
|
|
height=300
|
|||
|
|
)
|
|||
|
|
charts.append(('category', bar_chart, f'{col} 各类别的计数'))
|
|||
|
|
|
|||
|
|
# 3. 散点图（如果有多个数值列）
|
|||
|
|
if len(numeric_cols) >= 2:
|
|||
|
|
col1, col2 = numeric_cols[0], numeric_cols[1]
|
|||
|
|
color_col = cat_cols[0] if cat_cols and df[cat_cols[0]].nunique() <= 10 else None
|
|||
|
|
|
|||
|
|
scatter_encoding = {
|
|||
|
|
'x': alt.X(f'{col1}:Q', title=col1),
|
|||
|
|
'y': alt.Y(f'{col2}:Q', title=col2),
|
|||
|
|
'tooltip': [col1, col2]
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
if color_col:
|
|||
|
|
scatter_encoding['color'] = alt.Color(f'{color_col}:N', title=color_col)
|
|||
|
|
scatter_encoding['tooltip'].append(color_col)
|
|||
|
|
|
|||
|
|
scatter_chart = alt.Chart(df).mark_circle(
|
|||
|
|
size=60,
|
|||
|
|
opacity=0.6
|
|||
|
|
).encode(
|
|||
|
|
**scatter_encoding
|
|||
|
|
).properties(
|
|||
|
|
title=f'🔵 {col1} vs {col2} 散点图',
|
|||
|
|
height=350
|
|||
|
|
).interactive()
|
|||
|
|
|
|||
|
|
charts.append(('scatter', scatter_chart, f'{col1} 与 {col2} 的关系'))
|
|||
|
|
|
|||
|
|
# 4. 箱线图（数值列按分类）
|
|||
|
|
if numeric_cols and cat_cols:
|
|||
|
|
num_col = numeric_cols[0]
|
|||
|
|
cat_col = cat_cols[0]
|
|||
|
|
if df[cat_col].nunique() <= 10:
|
|||
|
|
box_chart = alt.Chart(df).mark_boxplot(
|
|||
|
|
color='#11998e'
|
|||
|
|
).encode(
|
|||
|
|
x=alt.X(f'{cat_col}:N', title=cat_col),
|
|||
|
|
y=alt.Y(f'{num_col}:Q', title=num_col),
|
|||
|
|
color=alt.Color(f'{cat_col}:N', legend=None)
|
|||
|
|
).properties(
|
|||
|
|
title=f'📦 {num_col} 按 {cat_col} 分组箱线图',
|
|||
|
|
height=300
|
|||
|
|
)
|
|||
|
|
charts.append(('boxplot', box_chart, f'{num_col} 在不同 {cat_col} 下的分布'))
|
|||
|
|
|
|||
|
|
# 5. 相关性热力图（如果有多个数值列）
|
|||
|
|
if len(numeric_cols) >= 3:
|
|||
|
|
corr_df = df[numeric_cols].corr().reset_index().melt(id_vars='index')
|
|||
|
|
corr_df.columns = ['var1', 'var2', 'correlation']
|
|||
|
|
|
|||
|
|
heatmap = alt.Chart(corr_df).mark_rect().encode(
|
|||
|
|
x=alt.X('var1:N', title=''),
|
|||
|
|
y=alt.Y('var2:N', title=''),
|
|||
|
|
color=alt.Color('correlation:Q',
|
|||
|
|
scale=alt.Scale(scheme='redblue', domain=[-1, 1]),
|
|||
|
|
title='相关系数'),
|
|||
|
|
tooltip=['var1', 'var2', alt.Tooltip('correlation:Q', format='.2f')]
|
|||
|
|
).properties(
|
|||
|
|
title='🔥 相关性热力图',
|
|||
|
|
height=300
|
|||
|
|
)
|
|||
|
|
charts.append(('heatmap', heatmap, '各数值变量间的相关性'))
|
|||
|
|
|
|||
|
|
return charts
|
|||
|
|
|
|||
|
|
|
|||
|
|
# ============== 初始化 Session State ==============
|
|||
|
|
|
|||
|
|
if "df" not in st.session_state:
|
|||
|
|
st.session_state.df = None
|
|||
|
|
|
|||
|
|
if "data_summary" not in st.session_state:
|
|||
|
|
st.session_state.data_summary = None
|
|||
|
|
|
|||
|
|
if "analyses" not in st.session_state:
|
|||
|
|
st.session_state.analyses = {}
|
|||
|
|
|
|||
|
|
if "qa_history" not in st.session_state:
|
|||
|
|
st.session_state.qa_history = []
|
|||
|
|
|
|||
|
|
# ============== 侧边栏 ==============
|
|||
|
|
|
|||
|
|
with st.sidebar:
|
|||
|
|
st.markdown("## 🔧 设置")
|
|||
|
|
|
|||
|
|
# API Key 输入
|
|||
|
|
api_key_input = st.text_input(
|
|||
|
|
"DeepSeek API Key",
|
|||
|
|
value=API_KEY,
|
|||
|
|
type="password",
|
|||
|
|
help="输入你的 DeepSeek API Key"
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
if api_key_input:
|
|||
|
|
API_KEY = api_key_input
|
|||
|
|
|
|||
|
|
st.divider()
|
|||
|
|
|
|||
|
|
# 文件上传
|
|||
|
|
st.markdown("## 📁 数据上传")
|
|||
|
|
|
|||
|
|
uploaded_file = st.file_uploader(
|
|||
|
|
"上传 CSV 文件",
|
|||
|
|
type=['csv'],
|
|||
|
|
help="支持 CSV 格式的数据文件"
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
# 或使用示例数据
|
|||
|
|
use_sample = st.checkbox("使用示例数据", value=False)
|
|||
|
|
|
|||
|
|
if use_sample:
|
|||
|
|
# 生成示例数据
|
|||
|
|
import numpy as np
|
|||
|
|
np.random.seed(42)
|
|||
|
|
|
|||
|
|
sample_df = pd.DataFrame({
|
|||
|
|
'姓名': [f'员工{i}' for i in range(1, 101)],
|
|||
|
|
'部门': np.random.choice(['技术部', '市场部', '财务部', '人事部', '运营部'], 100),
|
|||
|
|
'年龄': np.random.randint(22, 55, 100),
|
|||
|
|
'工龄': np.random.randint(1, 20, 100),
|
|||
|
|
'月薪': np.random.randint(8000, 50000, 100),
|
|||
|
|
'绩效评分': np.round(np.random.uniform(60, 100, 100), 1),
|
|||
|
|
'满意度': np.round(np.random.uniform(1, 5, 100), 1)
|
|||
|
|
})
|
|||
|
|
|
|||
|
|
st.session_state.df = sample_df
|
|||
|
|
st.session_state.data_summary = get_data_summary(sample_df)
|
|||
|
|
st.success("✅ 已加载示例数据")
|
|||
|
|
|
|||
|
|
elif uploaded_file is not None:
|
|||
|
|
try:
|
|||
|
|
df = pd.read_csv(uploaded_file)
|
|||
|
|
st.session_state.df = df
|
|||
|
|
st.session_state.data_summary = get_data_summary(df)
|
|||
|
|
st.session_state.analyses = {} # 重置分析结果
|
|||
|
|
st.session_state.qa_history = [] # 重置问答历史
|
|||
|
|
st.success(f"✅ 成功加载 {len(df)} 行数据")
|
|||
|
|
except Exception as e:
|
|||
|
|
st.error(f"❌ 文件加载失败: {str(e)}")
|
|||
|
|
|
|||
|
|
# 显示数据信息
|
|||
|
|
if st.session_state.df is not None:
|
|||
|
|
st.divider()
|
|||
|
|
st.markdown("### 📊 数据概览")
|
|||
|
|
df = st.session_state.df
|
|||
|
|
st.markdown(f"- **行数**: {len(df)}")
|
|||
|
|
st.markdown(f"- **列数**: {len(df.columns)}")
|
|||
|
|
st.markdown(f"- **数值列**: {len(df.select_dtypes(include=['number']).columns)}")
|
|||
|
|
st.markdown(f"- **分类列**: {len(df.select_dtypes(include=['object', 'category']).columns)}")
|
|||
|
|
|
|||
|
|
# ============== 主界面 ==============
|
|||
|
|
|
|||
|
|
st.markdown('<h1 class="main-title">📊 AI 数据分析师</h1>', unsafe_allow_html=True)
|
|||
|
|
st.markdown("*上传数据，让 AI 帮你发现数据中的故事*")
|
|||
|
|
|
|||
|
|
# 检查 API Key
|
|||
|
|
if not API_KEY:
|
|||
|
|
st.warning("⚠️ 请在侧边栏输入你的 DeepSeek API Key")
|
|||
|
|
st.stop()
|
|||
|
|
|
|||
|
|
# 检查数据
|
|||
|
|
if st.session_state.df is None:
|
|||
|
|
st.info("👈 请在侧边栏上传 CSV 文件或使用示例数据")
|
|||
|
|
|
|||
|
|
# 显示使用说明
|
|||
|
|
with st.expander("📖 使用说明", expanded=True):
|
|||
|
|
st.markdown("""
|
|||
|
|
### 欢迎使用 AI 数据分析师！
|
|||
|
|
|
|||
|
|
**功能介绍：**
|
|||
|
|
1. **📈 数据预览** - 查看数据基本信息和统计摘要
|
|||
|
|
2. **📊 智能可视化** - 自动生成适合数据的图表
|
|||
|
|
3. **🤖 AI 分析** - AI 深度分析数据特征和洞察
|
|||
|
|
4. **💬 数据问答** - 用自然语言询问关于数据的问题
|
|||
|
|
5. **📝 生成报告** - 一键生成完整的分析报告
|
|||
|
|
|
|||
|
|
**开始使用：**
|
|||
|
|
1. 在左侧边栏输入你的 DeepSeek API Key
|
|||
|
|
2. 上传 CSV 文件或勾选"使用示例数据"
|
|||
|
|
3. 切换不同标签页探索各项功能
|
|||
|
|
""")
|
|||
|
|
st.stop()
|
|||
|
|
|
|||
|
|
# 创建 OpenAI 客户端
|
|||
|
|
client = OpenAI(api_key=API_KEY, base_url=BASE_URL)
|
|||
|
|
df = st.session_state.df
|
|||
|
|
|
|||
|
|
# 创建标签页
|
|||
|
|
tab1, tab2, tab3, tab4, tab5 = st.tabs([
|
|||
|
|
"📈 数据预览",
|
|||
|
|
"📊 智能可视化",
|
|||
|
|
"🤖 AI 分析",
|
|||
|
|
"💬 数据问答",
|
|||
|
|
"📝 生成报告"
|
|||
|
|
])
|
|||
|
|
|
|||
|
|
# ============== Tab 1: 数据预览 ==============
|
|||
|
|
with tab1:
|
|||
|
|
st.header("📈 数据预览")
|
|||
|
|
|
|||
|
|
# 统计卡片
|
|||
|
|
col1, col2, col3, col4 = st.columns(4)
|
|||
|
|
|
|||
|
|
with col1:
|
|||
|
|
st.markdown(f"""
|
|||
|
|
<div class="stat-card">
|
|||
|
|
<h2>{len(df)}</h2>
|
|||
|
|
<p>数据行数</p>
|
|||
|
|
</div>
|
|||
|
|
""", unsafe_allow_html=True)
|
|||
|
|
|
|||
|
|
with col2:
|
|||
|
|
st.markdown(f"""
|
|||
|
|
<div class="stat-card">
|
|||
|
|
<h2>{len(df.columns)}</h2>
|
|||
|
|
<p>数据列数</p>
|
|||
|
|
</div>
|
|||
|
|
""", unsafe_allow_html=True)
|
|||
|
|
|
|||
|
|
with col3:
|
|||
|
|
missing = df.isnull().sum().sum()
|
|||
|
|
st.markdown(f"""
|
|||
|
|
<div class="stat-card">
|
|||
|
|
<h2>{missing}</h2>
|
|||
|
|
<p>缺失值数量</p>
|
|||
|
|
</div>
|
|||
|
|
""", unsafe_allow_html=True)
|
|||
|
|
|
|||
|
|
with col4:
|
|||
|
|
numeric_cols = len(df.select_dtypes(include=['number']).columns)
|
|||
|
|
st.markdown(f"""
|
|||
|
|
<div class="stat-card">
|
|||
|
|
<h2>{numeric_cols}</h2>
|
|||
|
|
<p>数值列数量</p>
|
|||
|
|
</div>
|
|||
|
|
""", unsafe_allow_html=True)
|
|||
|
|
|
|||
|
|
st.markdown("---")
|
|||
|
|
|
|||
|
|
# 数据表格
|
|||
|
|
st.subheader("📋 数据表格")
|
|||
|
|
st.dataframe(df, use_container_width=True, height=400)
|
|||
|
|
|
|||
|
|
# 统计摘要
|
|||
|
|
col_left, col_right = st.columns(2)
|
|||
|
|
|
|||
|
|
with col_left:
|
|||
|
|
st.subheader("📊 数值列统计")
|
|||
|
|
numeric_df = df.select_dtypes(include=['number'])
|
|||
|
|
if not numeric_df.empty:
|
|||
|
|
st.dataframe(numeric_df.describe(), use_container_width=True)
|
|||
|
|
else:
|
|||
|
|
st.info("没有数值列")
|
|||
|
|
|
|||
|
|
with col_right:
|
|||
|
|
st.subheader("📝 列信息")
|
|||
|
|
info_df = pd.DataFrame({
|
|||
|
|
'列名': df.columns,
|
|||
|
|
'数据类型': df.dtypes.astype(str).values,
|
|||
|
|
'非空值数': df.count().values,
|
|||
|
|
'唯一值数': [df[col].nunique() for col in df.columns]
|
|||
|
|
})
|
|||
|
|
st.dataframe(info_df, use_container_width=True, hide_index=True)
|
|||
|
|
|
|||
|
|
# ============== Tab 2: 智能可视化 ==============
|
|||
|
|
with tab2:
|
|||
|
|
st.header("📊 智能可视化")
|
|||
|
|
st.markdown("*根据数据特征自动生成适合的图表*")
|
|||
|
|
|
|||
|
|
charts = create_visualizations(df)
|
|||
|
|
|
|||
|
|
if not charts:
|
|||
|
|
st.warning("数据列类型不足以生成可视化图表")
|
|||
|
|
else:
|
|||
|
|
for i, (chart_type, chart, description) in enumerate(charts):
|
|||
|
|
st.markdown(f"**{description}**")
|
|||
|
|
st.altair_chart(chart, use_container_width=True)
|
|||
|
|
if i < len(charts) - 1:
|
|||
|
|
st.divider()
|
|||
|
|
|
|||
|
|
# ============== Tab 3: AI 分析 ==============
|
|||
|
|
with tab3:
|
|||
|
|
st.header("🤖 AI 智能分析")
|
|||
|
|
|
|||
|
|
analysis_type = st.selectbox(
|
|||
|
|
"选择分析类型",
|
|||
|
|
options=[
|
|||
|
|
("general", "🔍 综合分析 - 全面了解数据特征"),
|
|||
|
|
("correlation", "🔗 相关性分析 - 探索变量间关系"),
|
|||
|
|
("anomaly", "⚠️ 异常检测 - 发现数据问题")
|
|||
|
|
],
|
|||
|
|
format_func=lambda x: x[1]
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
if st.button("🚀 开始 AI 分析", type="primary", use_container_width=True):
|
|||
|
|
with st.spinner("AI 正在分析数据，请稍候..."):
|
|||
|
|
try:
|
|||
|
|
result = analyze_with_ai(
|
|||
|
|
client,
|
|||
|
|
st.session_state.data_summary,
|
|||
|
|
analysis_type[0]
|
|||
|
|
)
|
|||
|
|
st.session_state.analyses[analysis_type[0]] = result
|
|||
|
|
st.success("✅ 分析完成！")
|
|||
|
|
except Exception as e:
|
|||
|
|
st.error(f"❌ 分析出错: {str(e)}")
|
|||
|
|
|
|||
|
|
# 显示分析结果
|
|||
|
|
if analysis_type[0] in st.session_state.analyses:
|
|||
|
|
st.markdown("### 📋 分析结果")
|
|||
|
|
st.markdown(st.session_state.analyses[analysis_type[0]])
|
|||
|
|
|
|||
|
|
# 显示历史分析
|
|||
|
|
other_analyses = {k: v for k, v in st.session_state.analyses.items() if k != analysis_type[0]}
|
|||
|
|
if other_analyses:
|
|||
|
|
st.divider()
|
|||
|
|
with st.expander("📚 查看其他分析结果"):
|
|||
|
|
for key, value in other_analyses.items():
|
|||
|
|
type_names = {"general": "综合分析", "correlation": "相关性分析", "anomaly": "异常检测"}
|
|||
|
|
st.markdown(f"#### {type_names.get(key, key)}")
|
|||
|
|
st.markdown(value)
|
|||
|
|
st.divider()
|
|||
|
|
|
|||
|
|
# ============== Tab 4: 数据问答 ==============
|
|||
|
|
with tab4:
|
|||
|
|
st.header("💬 数据问答")
|
|||
|
|
st.markdown("*用自然语言询问关于数据的任何问题*")
|
|||
|
|
|
|||
|
|
# 示例问题
|
|||
|
|
with st.expander("💡 示例问题"):
|
|||
|
|
st.markdown("""
|
|||
|
|
- 这个数据集的主要特征是什么？
|
|||
|
|
- 哪个部门的平均薪资最高？
|
|||
|
|
- 年龄和工资之间有什么关系？
|
|||
|
|
- 数据中有哪些异常值？
|
|||
|
|
- 如何提高员工满意度？
|
|||
|
|
""")
|
|||
|
|
|
|||
|
|
# 显示对话历史
|
|||
|
|
for msg in st.session_state.qa_history:
|
|||
|
|
avatar = "👤" if msg["role"] == "user" else "🤖"
|
|||
|
|
with st.chat_message(msg["role"], avatar=avatar):
|
|||
|
|
st.markdown(msg["content"])
|
|||
|
|
|
|||
|
|
# 问题输入
|
|||
|
|
if question := st.chat_input("输入你关于数据的问题..."):
|
|||
|
|
# 显示用户问题
|
|||
|
|
with st.chat_message("user", avatar="👤"):
|
|||
|
|
st.markdown(question)
|
|||
|
|
|
|||
|
|
# AI 回答
|
|||
|
|
with st.chat_message("assistant", avatar="🤖"):
|
|||
|
|
with st.spinner("思考中..."):
|
|||
|
|
try:
|
|||
|
|
answer = ask_data_question(
|
|||
|
|
client,
|
|||
|
|
st.session_state.data_summary,
|
|||
|
|
question,
|
|||
|
|
st.session_state.qa_history
|
|||
|
|
)
|
|||
|
|
st.markdown(answer)
|
|||
|
|
|
|||
|
|
# 保存对话历史
|
|||
|
|
st.session_state.qa_history.append({"role": "user", "content": question})
|
|||
|
|
st.session_state.qa_history.append({"role": "assistant", "content": answer})
|
|||
|
|
|
|||
|
|
except Exception as e:
|
|||
|
|
st.error(f"❌ 回答出错: {str(e)}")
|
|||
|
|
|
|||
|
|
# 清空对话按钮
|
|||
|
|
if st.session_state.qa_history:
|
|||
|
|
if st.button("🗑️ 清空对话历史"):
|
|||
|
|
st.session_state.qa_history = []
|
|||
|
|
st.rerun()
|
|||
|
|
|
|||
|
|
# ============== Tab 5: 生成报告 ==============
|
|||
|
|
with tab5:
|
|||
|
|
st.header("📝 数据分析报告")
|
|||
|
|
|
|||
|
|
if not st.session_state.analyses:
|
|||
|
|
st.info("💡 请先在「AI 分析」标签页进行分析，然后再生成报告")
|
|||
|
|
else:
|
|||
|
|
st.markdown("*基于已有分析结果生成完整的数据分析报告*")
|
|||
|
|
|
|||
|
|
if st.button("📄 生成分析报告", type="primary", use_container_width=True):
|
|||
|
|
with st.spinner("正在生成报告，请稍候..."):
|
|||
|
|
try:
|
|||
|
|
report = generate_report(
|
|||
|
|
client,
|
|||
|
|
st.session_state.data_summary,
|
|||
|
|
st.session_state.analyses
|
|||
|
|
)
|
|||
|
|
st.session_state.report = report
|
|||
|
|
st.success("✅ 报告生成完成！")
|
|||
|
|
except Exception as e:
|
|||
|
|
st.error(f"❌ 生成报告出错: {str(e)}")
|
|||
|
|
|
|||
|
|
# 显示报告
|
|||
|
|
if "report" in st.session_state:
|
|||
|
|
st.markdown("---")
|
|||
|
|
st.markdown(st.session_state.report)
|
|||
|
|
|
|||
|
|
# 下载按钮
|
|||
|
|
st.download_button(
|
|||
|
|
label="📥 下载报告 (Markdown)",
|
|||
|
|
data=st.session_state.report,
|
|||
|
|
file_name="数据分析报告.md",
|
|||
|
|
mime="text/markdown"
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
# ============== 页脚 ==============
|
|||
|
|
st.divider()
|
|||
|
|
st.caption("Made with ❤️ using DeepSeek API + Streamlit | Python 程序设计课程设计示例")
|
|||
|
|
|