想在深度学习程序运行时动态存下来一些参数。
存成Excel文件查看方便,就查了几种方法,做个测试。因为我平常也不怎么用 Excel,简单的存取数据就够了。
xlwt/xlrd库 存Excel文件:(如果存储数据中有字符,那么写法还有点小小的变化)
import xlwt workbook = xlwt.Workbook(encoding='utf-8') booksheet = workbook.add_sheet('Sheet 1', cell_overwrite_ok=True) #存第一行cell(1,1)和cell(1,2) booksheet.write(0,0,34) booksheet.write(0,1,38) #存第二行cell(2,1)和cell(2,2) booksheet.write(1,0,36) booksheet.write(1,1,39) #存一行数据 rowdata = [43,56] for i in range(len(rowdata)): booksheet.write(2,i,rowdata[i]) workbook.save('test_xlwt.xls')
读Excel文件:(同样是对于数值类型数据)
import xlrd workbook = xlrd.open_workbook('D:\Py_exercise\test_xlwt.xls') print(workbook.sheet_names()) #查看所有sheet booksheet = workbook.sheet_by_index(0) #用索引取第一个sheet booksheet = workbook.sheet_by_name('Sheet 1') #或用名称取sheet #读单元格数据 cell_11 = booksheet.cell_value(0,0) cell_21 = booksheet.cell_value(1,0) #读一行数据 row_3 = booksheet.row_values(2) print(cell_11, cell_21, row_3) >>>34.0 36.0 [43.0, 56.0]
openpyxl 库 存Excel文件:
from openpyxl import Workbook workbook = Workbook() booksheet = workbook.active #获取当前活跃的sheet,默认是第一个sheet #存第一行单元格cell(1,1) booksheet.cell(1,1).value = 6 #这个方法索引从1开始 booksheet.cell("B1").value = 7 #存一行数据 booksheet.append([11,87]) workbook.save("test_openpyxl.xlsx")
读Excel文件:
from openpyxl import load_workbook workbook = load_workbook('D:\Py_exercise\test_openpyxl.xlsx') #booksheet = workbook.active #获取当前活跃的sheet,默认是第一个sheet sheets = workbook.get_sheet_names() #从名称获取sheet booksheet = workbook.get_sheet_by_name(sheets[0]) rows = booksheet.rows columns = booksheet.columns #迭代所有的行 for row in rows: line = [col.value for col in row] #通过坐标读取值 cell_11 = booksheet.cell('A1').value cell_11 = booksheet.cell(row=1, column=1).value
原理上其实都一样,就写法上有些差别。
其实如果对存储格式没有要求的话,我觉得存成 csv文件 也挺好的:
import pandas as pd csv_mat = np.empty((0,2),float) csv_mat = np.append(csv_mat, [[43,55]], axis=0) csv_mat = np.append(csv_mat, [[65,67]], axis=0) csv_pd = pd.DataFrame(csv_mat) csv_pd.to_csv("test_pd.csv", sep=',', header=False, index=False)
因为它读起来非常简单:
import pandas as pd filename = "D:\Py_exercise\test_pd.csv" csv_data = pd.read_csv(filename, header=None) csv_data = np.array(csv_data, dtype=float)