用Python做glmm可以吗？

Question

用Python做glmm (比如SPSS中的GENLINMIXED分析)可以吗？我是状态模型的忠实粉丝，但这个图书馆似乎不支持glmm.还有别的选择吗？

-编辑-

决定用R和r2py.

def RunAnalyseMLMlogit(dataset, outcomevars, meeneemvars, randintercept, randslope):

    from rpy2.robjects import pandas2ri
    from rpy2.robjects.packages import importr
    base = importr('base')
    stats = importr('stats')
    lme4 = importr('lme4')

    #data
    with SavReaderNp(dataset) as reader_np:
        array = reader_np.to_structured_array()

    df = pd.DataFrame(array)

    variabelen = ' '.join(outcomevars) + ' ~ ' + '+'.join(meeneemvars)
    randintercept2 = ['(1|'+i+')' for i in randintercept]
    intercept = '+'.join(randintercept2)
    randslope2 = ['(1+'+meeneemvars[0]+'|'+i+')' for i in randslope]
    slope = ' '.join(randslope2)

    pandas2ri.activate()
    r_df = pandas2ri.py2ri(df)

    #model
    #random intercepts + random slopes
    if len(randslope) > 0:
        formula = variabelen + '+' + intercept + '+' + slope

    #only random intercepts
    else:
        formula = variabelen + '+' + intercept

    model = lme4.glmer(formula, data=r_df, family= 'binomial')
    resultaat = base.summary(model).rx2('coefficients')
    uitkomst = base.summary(model)

    return uitkomst

Answer 1

根据this (诚然，不是最近的)文章，在Python中运行glmms仍然没有一个很好的解决方案。然而，如果你只是在寻找一个免费的(更灵活的！)替代在SPSS中运行测试，查看lme4包for R，您甚至可以使用一个包，例如rpy2，并直接从Python调用R，但这可能是一个小问题。