tfse-model-api-v0.2/Rating/scripts/financial_score.py

import pandas as pd

from Rating.db import find_threshold
from common.scripts import read_json_file


def financial_score(param1, param2):
    """
    财务要素进行打分
    Parameters:
        param1 string 二级行业
        param2 dict 财务指标
        weights dict 财务指标权重
        refers dict 行业阈值
        refer list 指标阈值
    Returns:
        scores dict 财务要素得分
    """
    # Parameters
    weights = read_json_file('/Rating/static/weights.json')
    refers = find_threshold(param1)[0]

    # Returns
    scores = dict()

    # main
    for key, value in param2.items():
        # 已获利息倍数为None，分数为满分
        if key == '已获利息倍数' and value is None:
            scores[key] = 5
        elif value is None:
            scores[key] = 0
        else:
            refer = list(map(float, refers[key]))
            weight = weights[key]
            standard = [weight, weight*0.8, weight*0.6, weight*0.4, weight*0.2]
            # 判断正反相关
            if refer[0] > refer[1]:
                if value >= refer[0]:
                    score = weight
                elif value > refer[1]:
                    score = linear_correlation_type(value, standard[0], standard[1], refer[0], refer[1])
                elif value > refer[2]:
                    score = linear_correlation_type(value, standard[1], standard[2], refer[1], refer[2])
                elif value > refer[3]:
                    score = linear_correlation_type(value, standard[2], standard[3], refer[2], refer[3])
                elif value > refer[4]:
                    score = linear_correlation_type(value, standard[3], standard[4], refer[3], refer[4])
                else:
                    score = 0
            else:
                if value <= refer[0]:
                    score = weight
                elif value < refer[1]:
                    score = linear_correlation_type(value, standard[0], standard[1], refer[0], refer[1])
                elif value < refer[2]:
                    score = linear_correlation_type(value, standard[1], standard[2], refer[1], refer[2])
                elif value < refer[3]:
                    score = linear_correlation_type(value, standard[2], standard[3], refer[2], refer[3])
                elif value < refer[4]:
                    score = linear_correlation_type(value, standard[3], standard[4], refer[3], refer[4])
                else:
                    score = 0
            scores[key] = round(score, 2)

    result = dict()

    fin1 = ['净资产收益率', '总资产报酬率']
    fin2 = ['总资产周转率', '应收账款周转率', '存货周转率']
    fin3 = ['资产负债率', '已获利息倍数', '速动比率']
    fin4 = ['营业增长率', '总资产增长率', '技术投入比率']

    result['盈利能力'] = dict((key, value) for key, value in scores.items() if key in fin1)
    result['资产质量'] = dict((key, value) for key, value in scores.items() if key in fin2)
    result['债务风险'] = dict((key, value) for key, value in scores.items() if key in fin3)
    result['经营增长'] = dict((key, value) for key, value in scores.items() if key in fin4)

    df_scores = pd.DataFrame([scores])
    result['盈利能力']['合计'] = round(float(df_scores[fin1].sum(axis=1).values[0]), 2)
    result['资产质量']['合计'] = round(float(df_scores[fin2].sum(axis=1).values[0]), 2)
    result['债务风险']['合计'] = round(float(df_scores[fin3].sum(axis=1).values[0]), 2)
    result['经营增长']['合计'] = round(float(df_scores[fin4].sum(axis=1).values[0]), 2)

    result['合计'] = round(sum(scores.values()), 2)

    return result


def linear_correlation_type(value, standard1, standard2, refer1, refer2):
    """
    线性相关类型 正相关/反相关
    Parameters
        value float 指标值
        standard1 float 标准分1
        standard2 float 标准分2
        refer1 阈值1
        refer2 阈值2
    Returns
        score float 得分
    """
    # main
    score = standard2 + (value - refer2) / (refer1 - refer2) * (standard1 - standard2)

    return score