Symbolic math for pandas.
Links symbolic expressions created using sympy to columns of data in pandas DataFrames. Any expression created with sympy, which can be "lambdified", will/should/might work.
The API design for sympa is subject to change, depending on feedback, if any.
from pandas import DataFrame
from sympy import symbols
from sympa import domath
#DataFrame expression:
df = pd.DataFrame({'x' : [1,2,3,4] * 2, 'y' : [0.1, 0.2] * 4})
#Series expression (Not recommended, but @kiaderouiche must have found it useful)
s = pd.Series({'x' : [1,2,3,4] * 2, 'y' : [0.1, 0.2] * 4})
# Notice x_-1 and x_-2 are used to reference x @ t=-1, and x @ t=-2.
x, xn1, xn2, y = symbols('x x_-1 x_-2 y')
f = 3.0 * x ** 3 + 2.0 * xn1 ** 2 + xn2 + (y * 2) / 2
df['f'] = domath(df,f)
For now, there isn't much documentation. I think it's pretty straightforward. Check out the /doc folder of the repo on github for more examples.
- Sympy >= 0.7.6 but should work with much older versions (untested).
- Pandas >= 0.15 but should work with much older versions (untested).
Works on 2.7, untested on all other versions. It is expected to be ported to 3.4 soon, and likely 2.6 as well.
pip install sympa