from src.data_insight import DataInsightFramework
import pandas as pd
import numpy as np

# 准备数据
data = pd.read_csv('your_data.csv')

# 创建框架实例
framework = DataInsightFramework()

# 一键分析
insights = framework.analyze(data, column='your_column')

# 查看结果
for insight in insights:
    print(f"[{insight.algorithm}] {insight.description}")

from src.data_insight import DataInsightFramework
import pandas as pd

# 创建框架实例
framework = DataInsightFramework()

# 分析DataFrame的某一列
df = pd.DataFrame({'sales': [100, 110, 105, 2000, 108]})  # 包含异常值
insights = framework.analyze(df, column='sales')

# 分析numpy数组
import numpy as np
data = np.random.randn(1000)
insights = framework.analyze(data)

# 自定义算法参数
config = {
    'zscore_threshold': 2.0,        # Z-Score阈值（默认2.5）
    'iqr_multiplier': 1.5,          # IQR倍数（默认1.5）
    'lof_neighbors': 20,            # LOF邻居数（默认20）
    'correlation_threshold': 0.7,    # 相关性阈值（默认0.7）
    'trend_alpha': 0.05             # 趋势检测显著性水平（默认0.05）
}

framework = DataInsightFramework(config=config)
insights = framework.analyze(data)

# 只运行特定算法
algorithms = ['zscore', 'iqr', 'mann_kendall']
insights = framework.analyze(data, algorithms=algorithms)

# 可选的算法名称
available_algorithms = [
    'basic_stats',      # 基础统计分析
    'zscore',          # Z-Score异常检测
    'iqr',             # IQR异常检测
    'lof',             # LOF局部异常检测
    'dbscan',          # DBSCAN聚类异常检测
    'mann_kendall',    # Mann-Kendall趋势检测
    'correlation',     # 相关性分析
    'cv_periodicity'   # 周期性检测
]

for insight in insights:
    print(f"算法: {insight.algorithm}")
    print(f"洞察类型: {insight.insight_type}")
    print(f"严重程度: {insight.severity}")  # low, medium, high, critical
    print(f"置信度: {insight.confidence}")   # 0-1之间
    print(f"描述: {insight.description}")
    print(f"建议: {insight.recommendations}")
    print(f"受影响的数据: {insight.affected_data}")  # 异常点索引
    print("---")

# 获取整体分析摘要
summary = framework.get_summary()
print(f"总洞察数: {summary['total_insights']}")
print(f"洞察类型分布: {summary['by_type']}")
print(f"严重程度分布: {summary['by_severity']}")
print(f"关键发现: {summary['key_findings']}")

# 导出为JSON格式
json_results = framework.export_results('json')
with open('insights.json', 'w', encoding='utf-8') as f:
    f.write(json_results)

# 导出为字典格式
dict_results = framework.export_results('dict')

# 监控销售数据异常
sales_data = pd.read_csv('daily_sales.csv')
framework = DataInsightFramework()
insights = framework.analyze(sales_data, column='revenue')

# 筛选高严重程度的异常
critical_insights = [i for i in insights if i.severity in ['high', 'critical']]

# 分析用户活跃度趋势
user_data = pd.read_csv('user_activity.csv')
insights = framework.analyze(user_data, column='daily_active_users')

# 查找趋势变化
trend_insights = [i for i in insights if i.insight_type == 'trend']

# 监控系统响应时间
perf_data = pd.read_csv('system_metrics.csv')
insights = framework.analyze(perf_data, column='response_time')

# 识别性能异常
anomaly_insights = [i for i in insights if i.insight_type == 'anomaly']

class DataInsightFramework:
    def _analyze_your_algorithm(self, data: np.ndarray):
        """添加您自己的算法"""
        # 实现您的算法逻辑
        result = your_algorithm(data)
        
        # 创建洞察结果
        insight = InsightResult(
            algorithm='your_algorithm',
            insight_type='your_type',
            severity='medium',
            confidence=0.8,
            description="您的算法描述",
            details={'your_details': result},
            recommendations=["您的建议"]
        )
        
        self.results.append(insight)