""" ███╗ ███╗ █████╗ ████████╗██╗ ██╗███████╗███╗ ███╗ █████╗ ████████╗██╗ ██████╗ █████╗ ██╗ ████╗ ████║██╔══██╗╚══██╔══╝██║ ██║██╔════╝████╗ ████║██╔══██╗╚══██╔══╝██║██╔════╝██╔══██╗██║ ██╔████╔██║███████║ ██║ ███████║█████╗ ██╔████╔██║███████║ ██║ ██║██║ ███████║██║ ██║╚██╔╝██║██╔══██║ ██║ ██╔══██║██╔══╝ ██║╚██╔╝██║██╔══██║ ██║ ██║██║ ██╔══██║██║ ██║ ╚═╝ ██║██║ ██║ ██║ ██║ ██║███████╗██║ ╚═╝ ██║██║ ██║ ██║ ██║╚██████╗██║ ██║███████╗ ╚═╝ ╚═╝╚═╝ ╚═╝ ╚═╝ ╚═╝ ╚═╝╚══════╝╚═╝ ╚═╝╚═╝ ╚═╝ ╚═╝ ╚═╝ ╚═════╝╚═╝ ╚═╝╚══════╝ ███╗ ███╗ ██████╗ ██████╗ ███████╗██╗ ██╗███╗ ██╗ ██████╗ ████╗ ████║██╔═══██╗██╔══██╗██╔════╝██║ ██║████╗ ██║██╔════╝ ██╔████╔██║██║ ██║██║ ██║█████╗ ██║ ██║██╔██╗ ██║██║ ███╗ ██║╚██╔╝██║██║ ██║██║ ██║██╔══╝ ██║ ██║██║╚██╗██║██║ ██║ ██║ ╚═╝ ██║╚██████╔╝██████╔╝███████╗███████╗██║██║ ╚████║╚██████╔╝ ╚═╝ ╚═╝ ╚═════╝ ╚═════╝ ╚══════╝╚══════╝╚═╝╚═╝ ╚═══╝ ╚═════╝ 团队成员: 王岩琪 袁欣萍 郭禹含 龚智勋 """ import pandas as pd import numpy as np def main(): """ 数据清洗 """ with pd.ExcelFile("/home/bobmaster/Downloads/数学建模/附件1：化学成分及力学性能.xlsx") as origin_data: pd_chemicals_raw = pd.read_excel(origin_data, "化学成分", usecols=[0, 2, 3, 4, 5, 6, 7]) pd_physics_raw = pd.read_excel(origin_data, "力学性能") pd_chemical = pd_chemicals_raw.iloc[1:, :] pd_physics = pd_physics_raw.dropna(how="any") # pd_chemical = pd_chemical.reindex(index = pd_chemical.index[::-1]) pd_physics_ronglianhao = pd_physics.iloc[:, 0].astype("int64") pd_physics_qufu = pd_physics.iloc[:, 2] pd_physics_kangla = pd_physics.iloc[:, 3] pd_physics_yanshen = pd_physics.iloc[:, 4] # 提取相同熔炼号的力学性能数据 comp_table = pd_physics.iloc[:, 0].duplicated(keep="last") # 比较表 # phy_num = pd_physics.count() # 力学表数据量 11213 phy_num = 11213 # phy_ronglianhao = [] phy_dict = {} phy_qufu = [] phy_kangla = [] phy_yanshen = [] temp = 0 for i in range(phy_num): phy_qufu.append(pd_physics_qufu[i]) phy_kangla.append(pd_physics_kangla[i]) phy_yanshen.append(pd_physics_yanshen[i]) if (comp_table[i] == False): # phy_ronglianhao[temp] = pd_physics_ronglianhao[i] phy_dict[pd_physics_ronglianhao[i]] = [phy_qufu, phy_kangla, phy_yanshen] temp += 1 phy_qufu = [] phy_kangla = [] phy_yanshen = [] # 数据规约 - 力学性能数据均值和标准差 phy_dict_qufu_mean = {} phy_dict_qufu_std = {} phy_dict_kangla_mean = {} phy_dict_kangla_std = {} phy_dict_yanshen_mean = {} phy_dict_yanshen_std = {} phy_dict_qufu_mean_list = [] phy_dict_qufu_std_list = [] phy_dict_kangla_mean_list = [] phy_dict_kangla_std_list = [] phy_dict_yanshen_mean_list = [] phy_dict_yanshen_std_list = [] for key in phy_dict: np_physics_array_qufu = np.array(phy_dict[key][0]) np_physics_array_kangla = np.array(phy_dict[key][1]) np_physics_array_yanshen = np.array(phy_dict[key][2]) phy_dict_qufu_mean[key] = np_physics_array_qufu.mean() phy_dict_qufu_std[key] = np_physics_array_qufu.std() phy_dict_kangla_mean[key] = np_physics_array_kangla.mean() phy_dict_kangla_std[key] = np_physics_array_kangla.std() phy_dict_yanshen_mean[key] = np_physics_array_yanshen.mean() phy_dict_yanshen_std[key] = np_physics_array_yanshen.std() # 清洗化学成分 # 重建索引保证在同一熔炼号的情况下与力学指标数据匹配 pd_chem_ronglianhao = pd_chemical.iloc[:, 0].astype("int64") pd_chem_ronglianhao = pd_chem_ronglianhao.drop_duplicates().reset_index().iloc[:, 1] pd_chem_E1_data = pd_chemical.iloc[:, 1].reset_index().iloc[:, 1] pd_chem_E2_data = pd_chemical.iloc[:, 2].reset_index().iloc[:, 1] pd_chem_E3_data = pd_chemical.iloc[:, 3].reset_index().iloc[:, 1] pd_chem_E4_data = pd_chemical.iloc[:, 4].reset_index().iloc[:, 1] pd_chem_E5_data = pd_chemical.iloc[:, 5].reset_index().iloc[:, 1] pd_chem_E6_data = pd_chemical.iloc[:, 6].reset_index().iloc[:, 1] pd_chem_E1 = {} pd_chem_E2 = {} pd_chem_E3 = {} pd_chem_E4 = {} pd_chem_E5 = {} pd_chem_E6 = {} temp = 0 # 数据规约 - 化学成分 # 0-701 清洗后得到的范围 for i in range(702): if (i % 2 != 0 and temp != 321): pd_chem_E1[pd_chem_ronglianhao[temp]] = (pd_chem_E1_data[i - 1] + pd_chem_E1_data[i]) / 2 pd_chem_E2[pd_chem_ronglianhao[temp]] = (pd_chem_E2_data[i - 1] + pd_chem_E2_data[i]) / 2 pd_chem_E3[pd_chem_ronglianhao[temp]] = (pd_chem_E3_data[i - 1] + pd_chem_E3_data[i]) / 2 pd_chem_E4[pd_chem_ronglianhao[temp]] = (pd_chem_E4_data[i - 1] + pd_chem_E4_data[i]) / 2 pd_chem_E5[pd_chem_ronglianhao[temp]] = (pd_chem_E5_data[i - 1] + pd_chem_E5_data[i]) / 2 pd_chem_E6[pd_chem_ronglianhao[temp]] = (pd_chem_E6_data[i - 1] + pd_chem_E6_data[i]) / 2 temp += 1 # 整理出最终所需数据并保证化学成分与力学性能数据一致性 E1_list = [] E2_list = [] E3_list = [] E4_list = [] E5_list = [] E6_list = [] for key in pd_chem_E1: if key in phy_dict: E1_list.append(pd_chem_E1[key]) E2_list.append(pd_chem_E2[key]) E3_list.append(pd_chem_E3[key]) E4_list.append(pd_chem_E4[key]) E5_list.append(pd_chem_E5[key]) E6_list.append(pd_chem_E6[key]) phy_dict_qufu_mean_list.append(phy_dict_qufu_mean[key]) phy_dict_qufu_std_list.append(phy_dict_qufu_std[key]) phy_dict_kangla_mean_list.append(phy_dict_kangla_mean[key]) phy_dict_kangla_std_list.append(phy_dict_kangla_std[key]) phy_dict_yanshen_mean_list.append(phy_dict_yanshen_mean[key]) phy_dict_yanshen_std_list.append(phy_dict_yanshen_std[key]) np_E1 = np.array(E1_list) np_E2 = np.array(E2_list) np_E3 = np.array(E3_list) np_E4 = np.array(E4_list) np_E5 = np.array(E5_list) np_E6 = np.array(E6_list) # 初始化二维直方图数据 # dist1 材料 dist1_E1 = np_E1 dist1_E2 = np_E2 dist1_E3 = np_E3 dist1_E4 = np_E4 dist1_E5 = np_E5 dist1_E6 = np_E6 # dist2 力学性能均值 dist2_qufu = np.array(phy_dict_qufu_mean_list) dist2_kangla = np.array(phy_dict_kangla_mean_list) dist2_yanshen = np.array(phy_dict_yanshen_mean_list) # dist3 力学性能标准差 dist3_qufu = np.array(phy_dict_qufu_std_list) dist3_kangla = np.array(phy_dict_kangla_std_list) dist3_yanshen = np.array(phy_dict_yanshen_std_list) # 绘制化学成分与力学特性关系的二维直方图 from hist2d import create_hist2d create_hist2d(dist1_E1, dist2_qufu, title="化学成分E1与屈服特性的关系", xlabel="E1 %", ylabel="屈服特性均值") create_hist2d(dist1_E1, dist2_kangla, title="化学成分E1与抗拉特性的关系", xlabel="E1 %", ylabel="抗拉特性均值") create_hist2d(dist1_E1, dist2_yanshen, title="化学成分E1与延伸率特性的关系", xlabel="E1 %", ylabel="延伸率特性均值") create_hist2d(dist1_E2, dist2_qufu, title="化学成分E2与屈服特性的关系", xlabel="E2 %", ylabel="屈服特性均值") create_hist2d(dist1_E2, dist2_kangla, title="化学成分E2与抗拉特性的关系", xlabel="E2 %", ylabel="抗拉特性均值") create_hist2d(dist1_E2, dist2_yanshen, title="化学成分E2与延伸率特性的关系", xlabel="E2 %", ylabel="延伸率特性均值") create_hist2d(dist1_E3, dist2_qufu, title="化学成分E3与屈服特性的关系", xlabel="E3 %", ylabel="屈服特性均值") create_hist2d(dist1_E3, dist2_kangla, title="化学成分E3与抗拉特性的关系", xlabel="E3 %", ylabel="抗拉特性均值") create_hist2d(dist1_E3, dist2_yanshen, title="化学成分E3与延伸率特性的关系", xlabel="E3 %", ylabel="延伸率特性均值") create_hist2d(dist1_E4, dist2_qufu, title="化学成分E4与屈服特性的关系", xlabel="E4 %", ylabel="屈服特性均值") create_hist2d(dist1_E4, dist2_kangla, title="化学成分E4与抗拉特性的关系", xlabel="E4 %", ylabel="抗拉特性均值") create_hist2d(dist1_E4, dist2_yanshen, title="化学成分E4与延伸率特性的关系", xlabel="E4 %", ylabel="延伸率特性均值") create_hist2d(dist1_E5, dist2_qufu, title="化学成分E5与屈服特性的关系", xlabel="E5 %", ylabel="屈服特性均值", fig=(10, 10)) create_hist2d(dist1_E5, dist2_kangla, title="化学成分E5与抗拉特性的关系", xlabel="E5 %", ylabel="抗拉特性均值", fig=(10, 10)) create_hist2d(dist1_E5, dist2_yanshen, title="化学成分E5与延伸率特性的关系", xlabel="E5 %", ylabel="延伸率特性均值", fig=(10, 10)) create_hist2d(dist1_E6, dist2_qufu, title="化学成分E6与屈服特性的关系", xlabel="E6 %", ylabel="屈服特性均值") create_hist2d(dist1_E6, dist2_kangla, title="化学成分E6与抗拉特性的关系", xlabel="E6 %", ylabel="抗拉特性均值") create_hist2d(dist1_E6, dist2_yanshen, title="化学成分E6与延伸率特性的关系", xlabel="E6 %", ylabel="延伸率特性均值") # 创建 OLS 回归模型 from linear_regression import OlsModel from linear_regression import ols_calcutate_all x = np.array([dist1_E1, dist1_E2, dist1_E3, dist1_E4, dist1_E5, dist1_E6]).transpose() # 材料与屈服特性均值回归模型 y = np.array(dist2_qufu) qufu_mean_ols_model = OlsModel(x, y) # 如需打印报告请删掉下一行的注释 # print(qufu_mean_ols_model.results.summary()) # 材料与抗拉特性均值回归模型 y = np.array(dist2_kangla) kangla_mean_ols_model = OlsModel(x, y) # 材料与延伸率特性均值回归模型 y = np.array(dist2_yanshen) yanshen_mean_ols_model = OlsModel(x, y) # 材料与屈服特性标准差回归模型 y = np.array(dist3_qufu) qufu_std_ols_model = OlsModel(x, y) # 材料与抗拉特性标准差回归模型 y = np.array(dist3_kangla) kangla_std_ols_model = OlsModel(x, y) # 材料与延伸率特性标准差回归模型 y = np.array(dist3_yanshen) yanshen_std_ols_model = OlsModel(x, y) # 给定熔炼号计算均值和标准差 # ronglianhao = 90624 # x1 = pd_chem_E1[ronglianhao] # x2 = pd_chem_E2[ronglianhao] # x3 = pd_chem_E3[ronglianhao] # x4 = pd_chem_E4[ronglianhao] # x5 = pd_chem_E5[ronglianhao] # x6 = pd_chem_E6[ronglianhao] # x = np.array([1, x1, x2, x3, x4, x5, x6]) # ols_calcutate_all(x, qufu_mean_ols_model, qufu_std_ols_model, # kangla_mean_ols_model, kangla_std_ols_model, # yanshen_mean_ols_model, yanshen_std_ols_model) """ 屈服均值: [281.04367017] 抗拉均值: [302.12712467] 延伸率均值: [11.72968023] 屈服标准差: [4.04484533] 抗拉标准差: [3.60625011] 延伸率标准差: [0.68357895] """ from linear_regression import MlrModel from linear_regression import mlr_calcutate_all # 创建 MLR 多元线性回归模型 x = np.array([dist1_E1, dist1_E2, dist1_E3, dist1_E4, dist1_E5, dist1_E6]).transpose() # 材料与屈服特性均值回归模型 y = np.array(dist2_qufu) qufu_mean_mlr_model = MlrModel(x, y) # 回归系数 # qufu_mean_mlr_model.results.coef_ # 常数，回归方程截距 # qufu_mean_mlr_model.results.intercept_ # 材料与抗拉特性均值回归模型 y = np.array(dist2_kangla) kangla_mean_mlr_model = MlrModel(x, y) # 材料与延伸率特性均值回归模型 y = np.array(dist2_yanshen) yanshen_mean_mlr_model = MlrModel(x, y) # 材料与屈服特性标准差回归模型 y = np.array(dist3_qufu) qufu_std_mlr_model = MlrModel(x, y) # 材料与抗拉特性标准差回归模型 y = np.array(dist3_kangla) kangla_std_mlr_model = MlrModel(x, y) # 材料与延伸率特性标准差回归模型 y = np.array(dist3_yanshen) yanshen_std_mlr_model = MlrModel(x, y) # 给定熔炼号计算均值和标准差 # ronglianhao = 90624 # x1 = pd_chem_E1[ronglianhao] # x2 = pd_chem_E2[ronglianhao] # x3 = pd_chem_E3[ronglianhao] # x4 = pd_chem_E4[ronglianhao] # x5 = pd_chem_E5[ronglianhao] # x6 = pd_chem_E6[ronglianhao] # x = np.array([1, x1, x2, x3, x4, x5, x6]).reshape(-1,6) # ols_calcutate_all(x, qufu_mean_ols_model, qufu_std_ols_model, # kangla_mean_ols_model, kangla_std_ols_model, # yanshen_mean_ols_model, yanshen_std_ols_model) """ 屈服均值: [281.04919773] 抗拉均值: [302.13923671] 延伸率均值: [11.75333675] 屈服标准差: [4.06391763] 抗拉标准差: [3.6079243] 延伸率标准差: [0.68167218] """ from heatmap import create_heatmap # 绘制热点图 dataset = pd.DataFrame( {'屈服': phy_dict_qufu_mean_list, '抗拉': phy_dict_kangla_mean_list, '延伸率': phy_dict_yanshen_mean_list, 'E1': np_E1, 'E2': np_E2, 'E3': np_E3, 'E4': np_E4, 'E5': np_E5, 'E6': np_E6}) create_heatmap(dataset) # from lineplot import create_lineplot # 绘制折线图 # x = "E1" 横坐标，从 E1-E6 选取 # y = "屈服" 纵坐标，从屈服、抗拉、延伸率 选取 # create_lineplot(dataset, x, y) if __name__=='__main__': main()