Python - Numerical Statistics

来源：互联网发布：sql数据库还原编辑：程序博客网时间：2024/05/18 22:42
# -*- coding: utf-8 -*-"""Created on Thu Apr 07 13:56:56 2016@author: wzhao3"""import jsonimport sysimport xlsxwriterimport globimport os#os.chdir("C:\\Users\\wzhao3\\Desktop\\CLV\\Shifu")#########################################def main(argv):    if len(argv) <1:        print "please input the stats target"        return             cat_set=[]    with open("categorical.config") as cat_file:    for line in cat_file:    cat_set.append(line.rstrip('\n'))      workbook = xlsxwriter.Workbook('Stat_numerical_Test.xlsx')    worksheets=[]    for arg in argv:        worksheets.append(workbook.add_worksheet(arg))        for ws_index, ws in enumerate(worksheets):                values=list()        variables=[]        header=['variable_name','sparkline']                for f_index, fileName in enumerate(sorted(glob.glob('files/ColumnConfig*.json'), key=lambda name: int(name.split('_')[1].split('.')[0]))):                        header.append(fileName.split('_')[1].split('.')[0])                with open(fileName) as data_file:                data = json.load(data_file)                content = []                for var in data:                    if var['columnName'] not in cat_set:                        if f_index==0:                            variables.append(var['columnName'])                        content.append(var['columnStats'][ws.get_name()])                values.append(content)                        header.append("STDEV")                  ws.write_row('A1', header)        ws.write_column('A2', variables)        ws.set_column('A:A', 20)        ws.set_column('B:B', 50)        values= map(list, zip(*values))            for i in range(len(values)):            ws.write_row('C'+str(i+2), values[i])        for ws in worksheets:        for i in range(len(variables)):            ws.add_sparkline('B'+str(i+2), {'range': ws.get_name()+'!C'+str(i+2)+':'+ col(len(header)-1) +str(i+2) })            ws.write_formula(col(len(header))+str(i+2), '{=STDEV('+ws.get_name()+'!C'+str(i+2)+':'+ col(len(header)-1) +str(i+2)+')}' )    workbook.close()def col(n):    if n<=26:        return chr(65+n-1)    if n>26 and n<=702:        c1=int(n/26)        c2=n%26        if c2==0:            return chr(65+c1-2)+chr(65+26-1)        else:            return chr(65+c1-1)+chr(65+c2-1)    if n>702 and n <= 17576:        c1=n%26%26        c2=int(n/26)        c3=c2%26        c4=int(c2/26)        if c1==0:            return  chr(65+c4-1)+chr(65+c3-2)+chr(65+26-1)        else:            return  chr(65+c4-1)+chr(65+c3-1)+chr(65+c1-1)        if __name__ == "__main__":    main(sys.argv[1:])
0 0