I'm using pandas dataframes and series as test and train data. I'm checking shapes of my train dataframe and test dataframe and they are absolutely identical. But I still have shapes not aligned error. Here is my fit/predict code:
train_df = df.loc[:50]
X_train = train_df[["Value", "Momentum", "Quality", "MinimumVolatility"]]
y_train = train_df["P1ExRe"]
X_train = sm.add_constant(X_train)
model = sm.OLS(y_train, X_train)
results = model.fit()
test_df = df.loc[51:100]
x_test = test_df[["Value", "Momentum", "Quality", "MinimumVolatility"]]
y_test = test_df["P1ExRe"]
print(x_test.shape==X_train.shape)
model.predict(x_test)
Here is the error:
ValueError Traceback (most recent call last)
<ipython-input-108-832ad1f6bc61> in <module>
4
5 print(x_test.shape==X_train.shape)
----> 6 model.predict(x_test)
~/projects/courserads/venv/lib/python3.6/site-packages/statsmodels/regression/linear_model.py in predict(self, params, exog)
378 exog = self.exog
379
--> 380 return np.dot(exog, params)
381
382 def get_distribution(self, params, scale, exog=None, dist_class=None):
<__array_function__ internals> in dot(*args, **kwargs)
ValueError: shapes (50,5) and (50,5) not aligned: 5 (dim 1) != 50 (dim 0)
You are using the model.predict method. You should use results.predict(...).
Model predict requires params, because only the results has the estimated parameters.
Your x_test in model.predict is interpreted as params and causes the shape mismatch.