Seeming Unrelated Regression weights and out of sample prediction

[tra6sdc]

Hello,

I am interested in performing seemingly unrelated regression using linearmodels. I would like to be able to weight the observations in the model and obtain out-of-sample predictions. Is this possible? I have adapted the example code to begin to explore the possibilities but not progressed and I don't know if this is because what I want to do isn't possible.

import pandas as pd
import numpy as np
data = pd.DataFrame(np.random.randn(500, 4), columns=['y1', 'x1_1', 'y2', 'x2_1'])
weight = pd.DataFrame(np.random.randn(500, 1), columns=['weight'])
from linearmodels.system import SUR
formula = {'eq1': 'y1 ~ 1 + x1_1', 'eq2': 'y2 ~ 1 + x2_1'}
mod = SUR.from_formula(formula, data, weights=weight)
res = mod.fit(cov_type='unadjusted')
res
#data1 = pd.DataFrame(np.random.randn(500, 4), columns=['y1', 'x1_1', 'y2', 'x2_1'])
#pre = res.predict(data1)

with error message:

C:\Program Files\Anaconda3\lib\site-packages\linearmodels\system\model.py:90: UserWarning: Weights not found for equation labels:
eq1, eq2
  warnings.warn(msg, UserWarning)

Thanks.

[bashtage]

The help says:

        weights : dict-like
            Dictionary like object (e.g. a DataFrame) containing variable
            weights.  Each entry must have the same number of observations as
            data.  If an equation label is not a key weights, the weights will
            be set to unity

so that weights should be a dictionary (or possibly a DataFrame) with columns eq1 and eq2 in your case where the column has the weights for respective weights. Let me know if this works.

[bashtage]

BTW, weights must be positive, so randn is a bad choice.

[tra6sdc]

Thanks for the help. Here is the code with adaptions that work in regards to weights.

import pandas as pd
import numpy as np
np.random.seed(123)
data = pd.DataFrame(np.random.randn(500, 4), columns=['y1', 'x1_1', 'y2', 'x2_1'])
weight = abs(pd.DataFrame(np.random.randn(500, 2), columns=['eq1','eq2']))
from linearmodels.system import SUR
formula = {'eq1': 'y1 ~ 1 + x1_1', 'eq2': 'y2 ~ 1 + x2_1'}
mod = SUR.from_formula(formula, data, weights=weight)
res = mod.fit(cov_type='unadjusted')
res

Is out-of-sample prediction possible? Or, for such a simple model should I do the calculation "manually"?

[bashtage]

Have a look at the help here:

https://bashtage.github.io/linearmodels/system/results.html#linearmodels.system.results.SystemResults.predict

[tra6sdc]

Here is my final code

import pandas as pd
import numpy as np
np.random.seed(123)
data = pd.DataFrame(np.random.randn(500, 4), columns=['y1', 'x1_1', 'y2', 'x2_1'])
weight = abs(pd.DataFrame(np.random.randn(500, 2), columns=['eq1','eq2']))
from linearmodels.system import SUR
formula = {'eq1': 'y1 ~ 1 + x1_1', 'eq2': 'y2 ~ 1 + x2_1'}
mod = SUR.from_formula(formula, data, weights=weight)
res = mod.fit(cov_type='unadjusted')
data1 = pd.DataFrame(np.random.randn(500, 4), columns=['y1', 'x1_1', 'y2', 'x2_1'])
pre = res.predict(data=data1)

Thanks again.