own_tech.py 文件源码

def load_file(file_name, sheet_name_list):
    '''
    load xlsx file into a dictionary indexed by sheet names
    :param string file_name:name of file
    :param [string] sheet_name_list: name of selected sheets in the xlsx file
    :return: {string:DataFrame} raw_data: {name of sheet:pure data retrieved from xlsx
    with column and index 0,1,2,...}
    '''
    print 'loading file...'
    cut_head = 2
    file = pd.ExcelFile(file_name)
    raw_data = {}
    # iterate over every sheet and retrieve useful data into raw_data
    for i in range(len(sheet_name_list)):
        print 'parsing sheet', sheet_name_list[i]
        # parse a sheet from the whole file into a DataFrame with headers cut off
        temp = file.parse(sheet_name_list[i]).iloc[cut_head:, :]
        # now temp.dtype = object,because the data read in contains string.Here convert it to float
        temp = temp.astype(np.float)
        # reset index and column with 0,1,2,...,
        temp.columns = range(temp.shape[1])
        temp.index = range(temp.shape[0])
        temp.fillna(0, inplace=True)
        raw_data[sheet_name_list[i]] = temp
    return raw_data