指标评估模型构建

新青年笔记

已于 2024-04-30 17:31:40 修改

阅读量169

点赞数 7

分类专栏：经验评估项目建设文章标签： python java

于 2024-04-30 13:54:40 首次发布

本文链接：https://blog.csdn.net/ANDYLAU_3S/article/details/138340656

版权

经验同时被 3 个专栏收录

2 篇文章 0 订阅

订阅专栏

项目建设

2 篇文章 0 订阅

订阅专栏

评估

1 篇文章 0 订阅

订阅专栏

参考

系统评价——指标数据的规范化处理（一） - 郝hai - 博客园 (cnblogs.com)

SPSSPRO-免费专业的在线数据分析平台

完整从0到1搭建指标体系 - 知乎 (zhihu.com)

构建多层级的目录结构指标体系，从低向上的聚合计算评估。

核心业务概念模型

核心业务流程

指标体系搭建

目录构建

简单实现构建规则，例如下面通过枚举类实现目录构建规则的定义，层级的设置，每个层级子级的设置，以及可分解的层级限制等规则。目的是构建一棵符合规则的树形菜单结构，构建结束，锁定目录，形成稳定的指标目录--版本记录。（锁定的时候可以对整棵树默认权重）

脑图模式：

权重构建

给指标目录每个层级赋值权重，赋值权重可以根据实际情况选择不同的权重计算方法。根据实际需要实现不同的权重计算算法。目的是对整棵树全部设置权重值。

算法：

下面举例权重分析算法-ahp层次分析法的实现：

java实现：

public class AHPMethod {
    public static JSONObject ahpMethod(double[][] dataset, String wd) {
        // 随机一致性指标R.I.（inc_rat）
        double[] inc_rat = {0, 0, 0, 0.58, 0.9, 1.12, 1.24, 1.32, 1.41, 1.45, 1.49, 1.51, 1.48, 1.56, 1.57, 1.59};
        // 数据深复制
        double[][] X = new double[dataset.length][dataset[0].length];
        for (int i = 0; i < dataset.length; i++) {
            X[i] = Arrays.copyOf(dataset[i], dataset[i].length);
        }
        // 生成权重向量
        double[] weights = new double[X[0].length];
        // 根据权重计算方法来确定各个权重
        if (wd.equals("m") || wd.equals("mean")) { // 均值
            for (int i = 0; i < X.length; i++) {
                double sum = 0;
                for (int j = 0; j < X[i].length; j++) {
                    int finalJ = j;
                    sum += X[i][j] / Arrays.stream(X).mapToDouble(row -> row[finalJ]).sum();
                }
                weights[i] = sum / X.length;
            }
        } else if (wd.equals("g") || wd.equals("geometric")) { // 几何
            for (int i = 0; i < X.length; i++) {
                double product = 1;
                for (double value : X[i]) {
                    product *= value;
                }
                weights[i] = Math.pow(product, 1.0 / X[i].length);
            }
            double sum = Arrays.stream(weights).sum();
            for (int i = 0; i < weights.length; i++) {
                weights[i] /= sum;
            }
        }
        // 计算特征根向量
        double[] vector = new double[X.length];
        for (int i = 0; i < X.length; i++) {
            double sum = 0;
            for (int j = 0; j < X[i].length; j++) {
                sum += X[i][j] * weights[j];
            }
            vector[i] = sum / weights[i];
        }
        // 获得平均特征根
        double lamb_max = Arrays.stream(vector).average().orElse(0);
        // 计算一致性指标
        double cons_ind = (lamb_max - X.length) / (X.length - 1);
        double ri = inc_rat[X.length];
        // 一致性判断
        double cr = cons_ind / ri;
        JSONObject data = new JSONObject();
        data.put("vector", vector);
        data.put("weights", weights);
        data.put("lambmax", lamb_max);
        data.put("ci", cons_ind);
        data.put("ri", ri);
        data.put("cr", cr);
        return data;
    }

python实现：

java调用python

public JSONObject ahpWeight(double[][] data, String flags) {
        flags = StringUtils.isEmpty(flags) ? "m" : flags;
        if (remoteSwitch) {
            try {
                JSONObject reqData = new JSONObject();
                reqData.put("data", data);
                reqData.put("flags", flags);
                log.info("reqData:{}", JSONObject.toJSONString(reqData));
                String res = HttpUtils.sendPost(remoteServerUrl + BizConstants.PYTHON_AHP_WEIGHT, JSONObject.toJSONString(reqData));
                log.info("resData:{}", res);
                return JSONObject.parseObject(res);
            } catch (Exception e) {
                throw new ServiceException("服务异常");
            }
        }

        //指定命令、路径、传递的参数
        String[] arguments = new String[]{"python",
                localPath + "/" + "ahp_weight.py",
                transforString(data),
                flags};
        return callPython(arguments);
    }

private JSONObject callPython(String[] arguments) {
        log.info("算法执行入参:{}", JSONObject.toJSONString(arguments));
        StringBuilder sbrs = null;
        StringBuilder sberror = null;
        try {
            ProcessBuilder builder = new ProcessBuilder(arguments);
            Process process = builder.start();
            BufferedReader in = new BufferedReader(new InputStreamReader(process.getInputStream(), "utf-8"));//获取字符输入流对象
            BufferedReader error = new BufferedReader(new InputStreamReader(process.getErrorStream(), "utf-8"));//获取错误信息的字符输入流对象
            String line = null;
            sbrs = new StringBuilder();
            sberror = new StringBuilder();
            //记录输出结果
            while ((line = in.readLine()) != null) {
                sbrs.append(line);
            }
            //记录错误信息
            while ((line = error.readLine()) != null) {
                sberror.append(line);
            }
            in.close();
            process.waitFor();
        } catch (Exception e) {
            e.printStackTrace();
        }
        if (sberror.length() > 0) {
            log.error("算法执行异常:{}", sberror);
            throw new ServiceException("算法执行异常：" + sberror);
        }
        log.info("算法执行结果:{}", JSONObject.toJSONString(sbrs));
        JSONObject result = JSONObject.parseObject(sbrs.toString());
        return result;
    }

本地python

import sys
import json
import ast
import numpy as np
from functools import reduce

def ahp_method(dataset, wd = 'm'):
    # 随机一致性指标R.I.（inc_rat）
    inc_rat  = np.array([0, 0, 0, 0.58, 0.9, 1.12, 1.24, 1.32, 1.41, 1.45, 1.49, 1.51, 1.48, 1.56, 1.57, 1.59])
    # 数据深复制
    X        = np.copy(dataset)
    # 生成权重向量
    weights  = np.zeros(X.shape[1])
    # 根据权重计算方法来确定各个权重
    if (wd == 'm' or wd == 'mean'):  #均值
        weights  = np.mean(X/np.sum(X, axis = 0), axis = 1)
    elif (wd == 'g' or wd == 'geometric'):  #几何
        for i in range (0, X.shape[1]):
            weights[i] = reduce( (lambda x, y: x * y), X[i,:])**(1/X.shape[1])
        weights = weights/np.sum(weights)      
    # 计算特征根向量
    vector   = np.sum(X*weights, axis = 1)/weights
    # 获得平均特征根
    lamb_max = np.mean(vector)
    # 计算一致性指标
    cons_ind = (lamb_max - X.shape[1])/(X.shape[1] - 1)
    ri=inc_rat[X.shape[1]]
    # 一致性判断
    cr       = cons_ind/ri
    return vector,weights,lamb_max,cons_ind,ri,cr

def create_json_response(vector,weights,lamb_max,cons_ind,ri,cr):
    data = {
        "vector": vector.tolist(),
        "weights": weights.tolist(),
        "lambmax": lamb_max,
        "ci": cons_ind,
        "ri": ri,
        "cr": cr,        
    }
    json_data = json.dumps(data)  # 将Python对象转换为JSON格式的字符串
    return json_data

# 从命令行参数中获取字符串表示的二维数组
dataset_str = sys.argv[1]
dataset = ast.literal_eval(dataset_str)
# 将Python列表转换为NumPy数组
dataset = np.array(dataset)

wd = sys.argv[2]
vector,weights,lamb_max,cons_ind,ri,cr = ahp_method(dataset, wd)
# weightslist=weights.tolist()
response = create_json_response(vector,weights,lamb_max,cons_ind,ri,cr)
print(response)

远程api接口（python）

@app.route('/ahp_weight', methods=['POST'])
def ahp_weight():    
    json_data = request.get_json()

    data = json_data.get('data')
    flags = json_data.get('flags')
    
    return aw.ahp_method(np.array(data),flags)  


import sys
import json
import ast
import numpy as np
from functools import reduce

def ahp_method(dataset, wd = 'm'):
    # 随机一致性指标R.I.（inc_rat）
    inc_rat  = np.array([0, 0, 0, 0.58, 0.9, 1.12, 1.24, 1.32, 1.41, 1.45, 1.49, 1.51, 1.48, 1.56, 1.57, 1.59])
    # 数据深复制
    X        = np.copy(dataset)
    # 生成权重向量
    weights  = np.zeros(X.shape[1])
    # 根据权重计算方法来确定各个权重
    if (wd == 'm' or wd == 'mean'):  #均值
        weights  = np.mean(X/np.sum(X, axis = 0), axis = 1)
    elif (wd == 'g' or wd == 'geometric'):  #几何
        for i in range (0, X.shape[1]):
            weights[i] = reduce( (lambda x, y: x * y), X[i,:])**(1/X.shape[1])
        weights = weights/np.sum(weights)      
    # 计算特征根向量
    vector   = np.sum(X*weights, axis = 1)/weights
    # 获得平均特征根
    lamb_max = np.mean(vector)
    # 计算一致性指标
    cons_ind = (lamb_max - X.shape[1])/(X.shape[1] - 1)
    ri=inc_rat[X.shape[1]]
    # 一致性判断
    cr       = cons_ind/ri
    data = {
        "vector": vector.tolist(),
        "weights": weights.tolist(),
        "lambmax": lamb_max,
        "ci": cons_ind,
        "ri": ri,
        "cr": cr,        
    }
    json_data = json.dumps(data)  # 将Python对象转换为JSON格式的字符串
    return json_data