Python函数定义时的参数解包：实现scipy curve_fit动态参数适配

算力阿峰

2026-10-12

解决方案

核心思路不需要动态修改函数定义的参数名，`scipy.optimize.curve_fit` 会按顺序传递待拟合参数，我们直接用可变参数`*args`接收后按规则拆分即可：

调整后的完整代码示例

import numpy as np
from scipy.optimize import curve_fit

def get_optimal_adstock_multivariate(x_var_names, final_df):
    y = np.array(final_df['Count Of Solutions'])
    n = len(x_var_names)
    # 生成参数名列表，后续用来映射最优结果
    coef_vars = [f"b{i+1}" for i in range(n)]
    rates = [f"rate_{i+1}" for i in range(n)]
    all_param_names = ["b0"] + coef_vars + rates

    def f(df, b0, *args):
        # args顺序和你定义的coef_vars + rates完全一致
        # 前n个为系数b，后n个为衰减率rate
        coefs = args[:n]
        rate_params = args[n:]
        
        # 这里替换为你自己的模型计算逻辑，以下是adstock模型示例
        pred = b0
        for idx in range(n):
            # 计算当前变量的adstock值
            col_vals = df[x_var_names[idx]]
            # 示例adstock计算：滞后加权求和
            adstock = 0
            adstock_vals = []
            for val in col_vals:
                adstock = val + rate_params[idx] * adstock
                adstock_vals.append(adstock)
            pred += coefs[idx] * np.array(adstock_vals)
        return pred

    # 调用curve_fit求解最优参数
    popt, pcov = curve_fit(f, final_df, y)
    # 把结果和参数名绑定，方便调用
    optimal_params = dict(zip(all_param_names, popt))
    return optimal_params, popt, pcov