panelexpr 0.1.0

A panel data evaluator.

panelexpr

A simple panel data evaluator

Installation

$ pip install panelexpr

Usage

Sample Data

sample_zh_2.csv:

Date	windcode	Open	High	Low	Close	Volume
20170913	000001.SZ	1221.488009	1226.803448	1205.54169	1215.109481	668237.42
20170914	000001.SZ	1215.109481	1232.118888	1194.910811	1203.415514	883087.77
20170915	000001.SZ	1200.22625	1203.415514	1185.34302	1200.22625	646094.81
20170918	000001.SZ	1195.973899	1203.415514	1190.658459	1195.973899	607612.96
20170919	000001.SZ	1195.973899	1205.54169	1177.901404	1183.216844	764212.62
20170913	000002.SZ	3473.691392	3576.666433	3431.128374	3536.849417	467463.77
20170914	000002.SZ	3539.595418	3852.639543	3508.016405	3783.989516	1150696.14
20170915	000002.SZ	3775.751513	4149.207662	3771.632511	4022.891612	1550495.61
20170918	000002.SZ	3983.074596	4083.303636	3843.028539	3849.893542	1063043.07
20170919	000002.SZ	3881.472555	4132.731656	3858.131546	3950.122582	1196308.5

load data

import pandas as pd
data = pd.read_csv("../data/sample_zh_2.csv")

Calculation

import panelexpr as pe
pe.eval("ma(Open, 2)", data=data, group_tag="windcode")

# equivalent to
data.groupby("windcode").rolling(2).mean()["Open"].reset_index(drop=True).rename()

Output:

0            NaN
1    1218.298745
2    1207.667866
3    1198.100075
4    1195.973899
5            NaN
6    3506.643405
7    3657.673466
8    3879.413054
9    3932.273575
dtype: float64

User defined operator

from panelexpr import TimeSeriesOperator

class MyMovingAverageOperator(TimeSeriesOperator):
    def eval(self, series: pd.Series, window) -> pd.Series: # all udo must implement this method
        s = series.rolling(window).mean()
        return s

pe.register("my_ma", MyMovingAverageOperator) # register the operator
pe.eval("my_ma(Open, 2)", data=data, group_tag="windcode")