我正在尝试分析具有多个回归（超过100个因子）的非常大的数据集。 绝大多数是虚拟变量/可以使用线性回归建模
但是，遵循三阶多项式，有没有一种方法可以轻松地在python中计算呢？我在Anaconda（v3.7）中使用了spyder。
我尝试了multipolyfit函数（对来自另一个线程的伪数据，链接如下），但无法正常工作，这是我的代码和错误。 Multivariate (polynomial) best fit curve in python?
from numpy import linalg, zeros, ones, hstack, asarray
import itertools
def basis_vector(n, i):
""" Return an array like [0, 0, ..., 1, ..., 0, 0]
>>> from multipolyfit.core import basis_vector
>>> basis_vector(3, 1)
array([0, 1, 0])
>>> basis_vector(5, 4)
array([0, 0, 0, 0, 1])
"""
x = zeros(n, dtype=int)
x[i] = 1
return x
def as_tall(x):
""" Turns a row vector into a column vector """
return x.reshape(x.shape + (1,))
def multipolyfit(xs, y, deg, full=False, model_out=False, powers_out=False):
"""
Least squares multivariate polynomial fit
Fit a polynomial like ``y = a**2 + 3a - 2ab + 4b**2 - 1``
with many covariates a, b, c, ...
Parameters
----------
xs : array_like, shape (M, k)
x-coordinates of the k covariates over the M sample points
y : array_like, shape(M,)
y-coordinates of the sample points.
deg : int
Degree o fthe fitting polynomial
model_out : bool (defaults to True)
If True return a callable function
If False return an array of coefficients
powers_out : bool (defaults to False)
Returns the meaning of each of the coefficients in the form of an
iterator that gives the powers over the inputs and 1
For example if xs corresponds to the covariates a,b,c then the array
[1, 2, 1, 0] corresponds to 1**1 * a**2 * b**1 * c**0
See Also
--------
numpy.polyfit
"""
y = asarray(y).squeeze()
rows = y.shape[0]
xs = asarray(xs)
num_covariates = xs.shape[1]
xs = hstack((ones((xs.shape[0], 1), dtype=xs.dtype) , xs))
generators = [basis_vector(num_covariates+1, i)
for i in range(num_covariates+1)]
# All combinations of degrees
powers = map(sum, itertools.combinations_with_replacement(generators, deg))
# Raise data to specified degree pattern, stack in order
A = hstack(asarray([as_tall((xs**p).prod(1)) for p in powers]))
beta = linalg.lstsq(A, y)[0]
if model_out:
return mk_model(beta, powers)
if powers_out:
return beta, powers
return beta
def mk_model(beta, powers):
""" Create a callable python function out of beta/powers from multipolyfit
This function is callable from within multipolyfit using the model_out flag
"""
# Create a function that takes in many x values
# and returns an approximate y value
def model(*args):
num_covariates = len(powers[0]) - 1
if len(args)!=(num_covariates):
raise ValueError("Expected %d inputs"%num_covariates)
xs = asarray((1,) + args)
return sum([coeff * (xs**p).prod()
for p, coeff in zip(powers, beta)])
return model
def mk_sympy_function(beta, powers):
from sympy import symbols, Add, Mul, S
num_covariates = len(powers[0]) - 1
xs = (S.One,) + symbols('x0:%d'%num_covariates)
return Add(*[coeff * Mul(*[x**deg for x, deg in zip(xs, power)])
for power, coeff in zip(powers, beta)])
错误：
************* Module untitled2
C0326: 49,56: : No space allowed before comma
xs = hstack((ones((xs.shape[0], 1), dtype=xs.dtype) , xs))
^
C0330: 52,0: : Wrong continued indentation (remove 10 spaces).
for i in range(num_covariates+1)]
| ^
C0326: 77,20: : Exactly one space required around comparison
if len(args)!=(num_covariates):
^^
C0330: 81,0: : Wrong continued indentation (remove 9 spaces).
for p, coeff in zip(powers, beta)])
| ^
C0330: 89,0: : Wrong continued indentation (remove 7 spaces).
for power, coeff in zip(powers, beta)])
| ^
C0303: 94,18: : Trailing whitespace
C0326: 96,10: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,13: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,16: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,19: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,22: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,25: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,29: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,32: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,36: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,39: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,43: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,47: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326: 96,51: : Exactly one space required after comma
data = [[1,1],[4,3],[8,3],[11,4],[10,7],[15,11],[16,12]]
^
C0326:100,26: : Exactly one space required after comma
stacked_x = numpy.array([x,x+1,x-1])
^
C0326:100,30: : Exactly one space required after comma
stacked_x = numpy.array([x,x+1,x-1])
^
C0303:101,31: : Trailing whitespace
C0304:104,0: : Final newline missing
C0114: 1,0: : Missing module docstring
C0103: 4,0: basis_vector: Argument name "n" doesn't conform to snake_case naming style
W0621: 12,4: basis_vector: Redefining name 'x' from outer scope (line 97)
C0103: 12,4: basis_vector: Variable name "x" doesn't conform to snake_case naming style
C0103: 16,0: as_tall: Argument name "x" doesn't conform to snake_case naming style
W0621: 16,12: as_tall: Redefining name 'x' from outer scope (line 97)
C0103: 20,0: multipolyfit: Argument name "xs" doesn't conform to snake_case naming style
C0103: 20,0: multipolyfit: Argument name "y" doesn't conform to snake_case naming style
W0621: 20,21: multipolyfit: Redefining name 'y' from outer scope (line 97)
R0913: 20,0: multipolyfit: Too many arguments (6/5)
C0103: 58,4: multipolyfit: Variable name "A" doesn't conform to snake_case naming style
W0613: 20,29: multipolyfit: Unused argument 'full'
W0612: 46,4: multipolyfit: Unused variable 'rows'
C0103: 79,8: mk_model.model: Variable name "xs" doesn't conform to snake_case naming style
C0116: 84,0: mk_sympy_function: Missing function or method docstring
C0103: 87,4: mk_sympy_function: Variable name "xs" doesn't conform to snake_case naming style
C0413: 92,0: : Import "import numpy" should be placed at the top of the module
C0413: 93,0: : Import "import matplotlib.pyplot as plt" should be placed at the top of the module
C0103: 94,0: : Constant name "mpf" doesn't conform to UPPER_CASE naming style
C0103: 96,0: : Constant name "data" doesn't conform to UPPER_CASE naming style
C0103: 97,0: : Constant name "x" doesn't conform to UPPER_CASE naming style
C0103: 97,3: : Constant name "y" doesn't conform to UPPER_CASE naming style
C0103:100,0: : Constant name "stacked_x" doesn't conform to UPPER_CASE naming style
C0103:101,0: : Constant name "coeffs" doesn't conform to UPPER_CASE naming style
E0602:101,27: : Undefined variable 'deg'
C0103:102,0: : Constant name "x2" doesn't conform to UPPER_CASE naming style
C0103:103,0: : Constant name "y2" doesn't conform to UPPER_CASE naming style
C0411: 2,0: : standard import "import itertools" should be placed before "from numpy import linalg, zeros, ones, hstack, asarray"
------------------------------------
Your code has been rated at -1.70/10'''