python - 回归汇总输出 : Order of categories

Question

这个问题是关于GLM的结果打印方式，也就是系数打印的顺序。我所说的“顺序”并不是指该术语的任何统计意义。

以下代码确定了一个线性模型:

from pandas import *
import statsmodels.api as sm
import patsy as patsy

df = read_csv("http://vincentarelbundock.github.io/Rdatasets/csv/ggplot2/diamonds.csv")

y, X = patsy.dmatrices( 'price ~  cut', data = df )

sm.GLM( y, X, family= sm.families.Gaussian()  ).fit().summary()

... 并生成以下输出，其中类别已排序:

(一般)、良好、理想、优质、非常好

====================================================================================
                       coef    std err          z      P>|z|      [95.0% Conf. Int.]
------------------------------------------------------------------------------------
Intercept         4358.7578     98.788     44.122      0.000      4165.137  4552.379
cut[T.Good]       -429.8933    113.849     -3.776      0.000      -653.034  -206.753
cut[T.Ideal]      -901.2158    102.412     -8.800      0.000     -1101.939  -700.493
cut[T.Premium]     225.4999    104.395      2.160      0.031        20.889   430.111
cut[T.Very Good]  -376.9979    105.164     -3.585      0.000      -583.116  -170.880
====================================================================================

我正在尝试做的事情:

我希望它们的顺序如下:

(一般)、好、非常好、优质、理想

我正在尝试做的事情在 R 中看起来像这样:

df = read.table( file = "http://vincentarelbundock.github.io/Rdatasets/csv/ggplot2/diamonds.csv",
            sep = ",",  header = TRUE)


df$cut = factor( df$cut, levels = c("Fair", "Good", "Very Good", "Premium", "Ideal"))

glm( price ~ cut, data = df, family = gaussian  )

注意输出中的排序遵循因​​子排序:

(一般)、好、非常好、优质、理想

Call:  glm(formula = price ~ cut, family = gaussian, data = df)

Coefficients:
 (Intercept)       cutGood  cutVery Good    cutPremium      cutIdeal  
      4358.8        -429.9        -377.0         225.5        -901.2

我如何在 Python 中执行此操作？

Answer 1

这是一个已知问题。我相信 PR 会受到欢迎。也许在这里继续对话？

https://github.com/pydata/patsy/issues/27