The optimization routine failed to converge: b'ABNORMAL_TERMINATION_IN_LNSRCH'

[tonytbui]

Hi, I ran cfa.fit() on a 33x119 dataset and got this error:

/Users/*/opt/anaconda3/lib/python3.7/site-packages/factor_analyzer/confirmatory_factor_analyzer.py:733: UserWarning: The optimization routine failed to converge: b'ABNORMAL_TERMINATION_IN_LNSRCH'
  'to converge: {}'.format(str(res.message)))
ConfirmatoryFactorAnalyzer(bounds=None, disp=True, impute='median',
                           is_cov_matrix=False, max_iter=200, n_obs=33,
                           specification=<factor_analyzer.confirmatory_factor_analyzer.ModelSpecification object at 0x1a1c2a7410>,
                           tol=None)

Here's the code I used:

model_dict = {"set_shifting": cued_cols+predict_cols,\ #15+15 columns
              "working_memory": DF_cols+nback_cols,\ #15+15 columns
              "attention": flanker_cols+shape_cols,\ #15+15 columns
              "response_inhibition": GNG_cols+stop_cols} #15+14 columns
model_spec = ModelSpecificationParser.parse_model_specification_from_dict(concat_mainDVs, model_dict)
cfa = ConfirmatoryFactorAnalyzer(model_spec, disp=True)
cfa.fit(concat_mainDVs.to_numpy())

[desilinguist]

It looks like the minimization that is done as part of factor model fitting does not converge for your specific dataset. Can you replicate the issue using R?

[tonytbui]

It looks like the minimization that is done as part of factor model fitting does not converge for your specific dataset. Can you replicate the issue using R?

Is there a setting that I can change in /Users/*/opt/anaconda3/lib/python3.7/site-packages/factor_analyzer/confirmatory_factor_analyzer.py to work around that minimization?

Also, by "using R," do you mean the Lavaan pakage?

[desilinguist]

No, there’s no such setting because the minimization is a core part of the model fitting.

Yes, R + lavaan is a reasonable counterpart.

[tonytbui]

No, there’s no such setting because the minimization is a core part of the model fitting.

Yes, R + lavaan is a reasonable counterpart.

I used lavaan and got somewhat more understandable errors:

Error in lav_samplestats_icov(COV = cov[[g]], ridge = ridge, x.idx = x.idx[[g]],  : 
  lavaan ERROR: sample covariance matrix is not positive-definite
In addition: Warning messages:
1: In lav_data_full(data = data, group = group, cluster = cluster,  :
  lavaan WARNING: some observed variances are (at least) a factor 1000 times larger than others; use varTable(fit) to investigate
2: In lav_data_full(data = data, group = group, cluster = cluster,  :
  lavaan WARNING: small number of observations (nobs < nvar)
  nobs = 33 nvar = 119

Update: I cut down to 2 indicators per factor and 4 factors in total. This avoided the (nobs < nvar) error but showed another error that might corroborate with python's factor_analyzer:

Warning messages:
1: In lav_data_full(data = data, group = group, cluster = cluster,  :
  lavaan WARNING: some observed variances are (at least) a factor 1000 times larger than others; use varTable(fit) to investigate
2: In lav_model_estimate(lavmodel = lavmodel, lavpartable = lavpartable,  :
  lavaan WARNING: the optimizer warns that a solution has NOT been found!