人工智能项目基于深度学习LSTM的原油与化工期货预测分析系统python源码+数据集+模型（毕业设计）.zip

import os import numpy as np import matplotlib.pyplot as plt import pandas as pd import seaborn as sns import xlwt import tensorflow as tf from PIL._imaging import display from numpy import mean from sklearn.preprocessing import StandardScaler from sklearn.model_selection import StratifiedShuffleSplit, train_test_split from pandas._testing import assert_frame_equal from sklearn.experimental import enable_iterative_imputer from sklearn.impute import IterativeImputer from sklearn.impute import KNNImputer # 不加这个无法绘图 plt.rcParams['font.sans-serif'] = ['KaiTi', 'SimHei', 'FangSong'] # 汉字字体,优先使用楷体，如果找不到楷体，则使用黑体 plt.rcParams['font.size'] = 12 # 字体大小 plt.rcParams['axes.unicode_minus'] = False # 正常显示负号 #读取xls文件 def load_data(): # 读取文件 df = pd.read_excel(r'..\data\期货价格（多变量）.xls', index_col='日期') return df #写入xls文件 def write_data(data): # 写入文件 book = xlwt.Workbook(encoding='utf-8', style_compression=0) x = pd.DataFrame(data) x.to_excel(r'F:\QH\Data\College\My Race\SRT\futures_prediction\data\汇总（多变量）.xls', sheet_name='多变量') savepath = r'F:\QH\Data\College\My Race\SRT\futures_prediction\data\汇总（多变量）.xls' book.save(savepath) #填补期货价格 def fill_price(df): # 多变量填补法 II=IterativeImputer(max_iter=10,random_state=0) df=II.fit_transform(df) # KNN填补法 # KI=KNNImputer(n_neighbors=10,weights="uniform") # price=KI.fit_transform(df) # 计算每月日均值 # for k in range(1, 32): # 每月31天 # df = df.replace(np.nan, 0) # # 初始化列表,用来存储一个月每天的价格均值 # day1 = [0 for j in range(28)] # i = 0 # i表示列数 # for index, col in df.iteritems(): # num = 0 # count = 0 # for index, row in df.iterrows(): # if row['DAY'] == k: # if row[i] != 0: # 有初始值的项 # num += row[i] # count += 1 # day1[i - 1] = num / count # i += 1 # print(day1) # df = df.replace(0, np.nan) # # 填补 # i = 1 # 列标 # for index, col in df.iteritems(): # j = 1 # 行标 # for index, row in df.iterrows(): # if row['DAY'] == k: # if i >= 29: # 退出循环，防止数组越界 # break # if j >= 3172: # break # if pd.isnull(row[i]): # df.iat[j - 1, i] = day1[i - 1] # 将求出的均值填充进df # j += 1 # i += 1 # print('第%d天完成' % (k)) return df #填补开工率 def fill_rate(df): # df = df.fillna(method='pad') # 计算每月日均值 for k in range(1, 32): # 每月31天 df = df.replace(np.nan, 0) # 初始化列表,用来存储一个月每天的价格均值 day1 = [0 for j in range(36)] i = 0 # i表示列数 for index, col in df.iteritems(): num = 0 count = 0 for index, row in df.iterrows(): if row['DAY'] == k: if row[i] != 0: # 有初始值的项 num += row[i] count += 1 day1[i - 1] = num / count i += 1 print(day1) df = df.replace(0, np.nan) # 填补 i = 1 # 列标 for index, col in df.iteritems(): j = 1 # 行标 for index, row in df.iterrows(): if row['DAY'] == k: if i >= 37: # 退出循环，防止数组越界 break # if j>=3172: # break if pd.isnull(row[i]): df.iat[j - 1, i] = day1[i - 1] # 将求出的均值填充进df j += 1 i += 1 print('第%d天完成' % (k)) return df #Z-score标准化 def stardand(df): scaler = StandardScaler() scaler.fit(df) df = scaler.transform(df) return df #PCA降维与重构 def pca(dataMat, topNfeat=27): # m=3171,n=64,p=23 meanVals = np.mean(dataMat, axis=0) # 均值：1*n meanRemoved = dataMat - meanVals # 去除均值 m*n covMat = np.cov(meanRemoved, rowvar=0) # 协方差矩阵 n*n eigVals, eigVects = np.linalg.eig(np.mat(covMat)) # 特征值 1*n，特征矩阵 n*n # for i in range(63):#每个柱子是前面所有柱子的累加 # eigVals[i+1]+=eigVals[i] # sum=eigVals[63] # for i in range(64):#求百分比 # eigVals[i]/=sum # index = np.arange(len(eigVals)); # plt.bar(index, eigVals, width=0.6) # plt.xticks([x for x in range(64) if x % 2 == 0]) # x标记step设置为2 # for a, b in zip(index, eigVals): # 柱子上的数字显示 # plt.text(a, b, '%.2f' % b, ha='center', va='bottom', fontsize=4); # plt.show() eigValInd = np.argsort(eigVals) # 将特征值从小到大索引排序 eigValInd = eigValInd[:-(topNfeat + 1):-1] # 从后往前仅保留p个列（将topNfeat理解为p即可） redEigVects = pd.DataFrame(eigVects[:, eigValInd]) # 仅保留p个最大特征值对应的特征向量，按从大到小的顺序重组特征矩阵n*p lowDDataMat = np.dot(meanRemoved, redEigVects) # 将数据转换到低维空间lowDDataMat： m*p reconMat = np.dot(pd.DataFrame(lowDDataMat), redEigVects.T) + meanRemoved # 从压缩空间重构原数据reconMat： m*n return lowDDataMat, reconMat if __name__ == '__main__': prall = load_data() # pd.set_option('display.max_rows', None) # 可以填数字，填None表示'行'无限制 # pd.set_option('display.max_columns', None) # 可以填数字，填None表示'列'无限制 df=stardand(prall) write_data(df) # write_data(df) # #PTA与原油 # plt.figure(figsize=(8,4)) # sns.lineplot(x='现货价:原油:英国布伦特Dtd',y='期货收盘价(活跃合约):精对苯二甲酸(PTA)',data=df) # plt.show() # #乙二醇与原油 # sns.lineplot(x='现货价:原油:英国布伦特Dtd', y='期货收盘价(活跃合约):乙二醇', data=df) # plt.show() # #短纤与原油 # sns.lineplot(x='现货价:原油:英国布伦特Dtd',y='期货收盘价(活跃合约):短纤',data=df) # plt.show()