python：np.diff() 计算差值和升跌幅百分比

计算基金净值的差值和升跌幅百分比，用于哈奇计划。

基金净值文件数据格式: 

date,jz,ljjz
2016-01-04,1.1141,1.1141
2016-01-05,1.1161,1.1161
2016-01-06,1.1350,1.1350

csv2txt.py

# coding=utf-8
import os, sys
import datetime
import numpy as np
import pandas as pd

if len(sys.argv) ==2:
    fcode = sys.argv[1]
else:
    print('usage: csv2txt.py fundcode ')
    sys.exit(1)

f1 = fcode +'.csv'
if not os.path.exists(f1):
    print("Error: %s not found." % f1)
    sys.exit(1)

fn,ext = os.path.splitext(f1)
if len(fn) !=6:
    print('Error: len(%s) !=6' % fn)
    sys.exit(1)
if ext !='.csv':
    print('Error: %s is not .csv' % f1)
    sys.exit(1)

df = pd.read_csv(f1, index_col='date')
if len(df) <5:
    print(" len(df) <5 ")
    sys.exit(2)

# 基金净值
net_values = np.array(df['jz'].values)
# Take diff of net values and computing rate of change
diff = np.diff(net_values)
diff_percentage = 100.0 * np.diff(net_values) / net_values[:-1]
# 四舍五入到所需精度的值
diff = np.around(diff,4)
diff_percentage = np.around(diff_percentage,4)

# DataFrame 新增2列: diff,p_change
df['diff'] = np.insert(diff,0,[0.0])
df['p_change'] = np.insert(diff_percentage,0,[0.0])
# export
df.to_csv(fcode+'.txt')

运行 python csv2txt.py 660008

结果文件数据格式: 

date,jz,ljjz,diff,p_change
2016-01-04,1.1141,1.1141,0.0,0.0
2016-01-05,1.1161,1.1161,0.002,0.1795
2016-01-06,1.135,1.135,0.0189,1.6934

参考书：[ Python机器学习经典实例 ] 第8章 解剖时间序列和时序数据