如何评价2024华为杯研究生数学建模竞赛（研赛）B题？

# 导入必要的库 import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns # 用于模型建立和评估的库 from sklearn.model_selection import train_test_split from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error, r2_score # 生成示例数据 np.random.seed(42) # 假设有1000条数据 num_samples = 1000 # 输入特征 data = pd.DataFrame({ 'RSSI_AP1_AP2': np.random.uniform(-90, -50, num_samples), # AP1与AP2之间的RSSI 'PD_threshold': np.full(num_samples, -82), # PD门限，固定值 'ED_threshold': np.full(num_samples, -62), # ED门限，固定值 'NAV_threshold': np.full(num_samples, -82), # NAV门限，固定值 'Traffic_Type': np.random.choice(['UDP', 'TCP'], num_samples), # 业务流量类型 'Packet_Length': np.random.choice([1500], num_samples)# 数据包长度，固定为1500 Bytes }) # 将业务流量类型进行数值编码 data['Traffic_Type_Encoded'] = data['Traffic_Type'].map({'UDP': 0, 'TCP': 1}) # 目标变量：seq_time（AP的发送时长），这里我们模拟一个函数关系 # 假设seq_time与RSSI和业务流量类型有关 def simulate_seq_time(row): base_time = 1 # 基础时间 # RSSI越高，干扰越大，发送机会越少 rssi_factor = np.exp((row['RSSI_AP1_AP2'] + 90) / 10) # TCP流量由于需要ACK，发送时长稍长 traffic_factor = 1.2 if row['Traffic_Type'] == 'TCP' else 1.0 # 模拟seq_time seq_time = base_time / rssi_factor * traffic_factor + np.random.normal(0, 0.1) return seq_time data['seq_time'] = data.apply(simulate_seq_time, axis=1) # 数据预览 print(data.head())

RSSI_AP1_AP2 PD_threshold ED_threshold NAV_threshold Traffic_Type \ 0 -70.163482 -82 -62 -82UDP 1 -86.816917 -82 -62 -82UDP 2 -82.203600 -82 -62 -82TCP 3 -77.820934 -82 -62 -82UDP 4 -71.674645 -82 -62 -82TCP Packet_Length Traffic_Type_Encoded seq_time 0 1500 0 0.702266 1 1500 0 1.333558 2 1500 1 1.430837 3 1500 0 0.979389 4 1500 1 1.111231

# 导入必要的库 import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns # 用于模型建立和评估的库 from sklearn.model_selection import train_test_split from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error, r2_score # 计算相关性 correlation = data[['RSSI_AP1_AP2', 'Traffic_Type_Encoded', 'seq_time']].corr() print("相关性矩阵：\n", correlation) # 可视化相关性 sns.heatmap(correlation, annot=True) plt.title('Feature Correlation') plt.show()

相关性矩阵： RSSI_AP1_AP2 Traffic_Type_Encoded seq_time RSSI_AP1_AP2 1.000000 -0.007921 -0.885037 Traffic_Type_Encoded-0.007921 1.000000 0.239820 seq_time -0.885037 0.239820 1.000000

# 导入必要的库 import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns # 用于模型建立和评估的库 from sklearn.model_selection import train_test_split from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error, r2_score # 选择特征和目标变量 X = data[['RSSI_AP1_AP2', 'Traffic_Type_Encoded']] y = data['seq_time'] # 拆分训练集和测试集 # 省略部分代码 # 建立线性回归模型 model = LinearRegression() model.fit(X_train, y_train) # 输出模型参数 print("模型系数：", model.coef_) print("模型截距：", model.intercept_)

模型系数： [-0.04127035 0.11639739] 模型截距： 3.894509739865504

# 导入必要的库 import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns # 用于模型建立和评估的库 from sklearn.model_selection import train_test_split from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error, r2_score # 在测试集上进行预测 y_pred = model.predict(X_test) # 计算均方误差和决定系数 mse = mean_squared_error(y_test, y_pred) r2 = r2_score(y_test, y_pred) print("均方误差（MSE）：", mse) print("决定系数（R^2）：", r2)

均方误差（MSE）： 0.009954319215601865 决定系数（R^2）： 0.7849128309137948

# 导入必要的库 import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns # 用于模型建立和评估的库 from sklearn.model_selection import train_test_split from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error, r2_score # 生成测试集数据 test_set_1 = pd.DataFrame({ 'RSSI_AP1_AP2': np.random.uniform(-90, -50, 200), 'Traffic_Type': np.random.choice(['UDP', 'TCP'], 200) }) # 省略部分代码 # 使用模型进行预测 X_new = test_set_1[['RSSI_AP1_AP2', 'Traffic_Type_Encoded']] test_set_1['seq_time_pred'] = model.predict(X_new) # 显示预测结果 print(test_set_1.head())

# 导入必要的库 import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns # 用于模型建立和评估的库 from sklearn.model_selection import train_test_split from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error, r2_score # 生成示例数据 num_samples = 1000 data2 = pd.DataFrame({ 'Signal_RSSI': np.random.uniform(-80, -40, num_samples), # 信号功率 'Interference_RSSI': np.random.uniform(-90, -50, num_samples), # 干扰功率 'Transmission_Type': np.random.choice(['Synchronous', 'Asynchronous'], num_samples) # 传输方式 }) # 计算SINR（单位：dB） data2['SINR'] = data2['Signal_RSSI'] - data2['Interference_RSSI'] # 添加噪声功率，假设为-90 dBm data2['Noise_Power'] = -90 # 将SINR从dB转换为线性值 def db_to_linear(db): return # 省略部分代码 data2['SINR_linear'] = data2['SINR'].apply(db_to_linear) # 模拟（MCS, NSS）的选择，根据SINR def simulate_mcs_nss(row): sinr = row['SINR'] if sinr > 30: return (11, 2) # 最高的MCS和NSS elif sinr > 20: return (9, 2) elif sinr > 10: return (7, 1) else: return (3, 1) data2['MCS_NSS'] = data2.apply(simulate_mcs_nss, axis=1) data2[['MCS', 'NSS']] = pd.DataFrame(data2['MCS_NSS'].tolist(), index=data2.index) # 数据预览 print(data2.head())

Signal_RSSI Interference_RSSI Transmission_Type SINR Noise_Power \ 0 -51.715751 -76.481486 Synchronous 24.765735-90 1 -77.439573 -67.715012 Asynchronous -9.724561-90 2 -56.089729 -81.546503 Synchronous 25.456773-90 3 -65.416353 -77.017649 Synchronous 11.601296-90 4 -49.240923 -57.271143 Synchronous 8.030220-90 SINR_linear MCS_NSS MCS NSS 0 298.471848 (9, 2) 9 2 1 0.106852 (3, 1) 3 1 2 350.283628 (9, 2) 9 2 3 14.451053 (7, 1) 7 1 4 6.350072 (3, 1) 3 1

# 导入必要的库 import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns # 用于模型建立和评估的库 from sklearn.model_selection import train_test_split from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error, r2_score # 将传输方式编码 data2['Transmission_Type_Encoded'] = data2['Transmission_Type'].map({'Synchronous': 0, 'Asynchronous': 1}) # 特征和目标变量 X = data2[['SINR', 'Transmission_Type_Encoded']] y = data2['MCS'] # 拆分训练集和测试集 X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42) # 使用决策树分类模型 from sklearn.tree import DecisionTreeClassifier # 省略部分代码 # 预测 y_pred = classifier.predict(X_test) # 评估模型 from sklearn.metrics import accuracy_score, classification_report accuracy = accuracy_score(y_test, y_pred) print("模型准确率：", accuracy) print("分类报告：\n", classification_report(y_test, y_pred))

# 导入必要的库 import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns # 用于模型建立和评估的库 from sklearn.model_selection import train_test_split from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error, r2_score # 生成测试集数据 test_set_2 = pd.DataFrame({ 'Signal_RSSI': np.random.uniform(-80, -40, 200), 'Interference_RSSI': np.random.uniform(-90, -50, 200), 'Transmission_Type': np.random.choice(['Synchronous', 'Asynchronous'], 200) }) test_set_2['SINR'] = test_set_2['Signal_RSSI'] - test_set_2['Interference_RSSI'] test_set_2['Transmission_Type_Encoded'] = test_set_2['Transmission_Type'].map({'Synchronous': 0, 'Asynchronous': 1}) # 预测MCS # 省略部分代码 # 显示预测结果 print(test_set_2.head())