AttributeError: DataFrame object has no attribute 'date'

Question

I am doing time series analysis using statsmodels in Python. I am getting an error while using ARIMA method. I can select the parameter for order and seasonal_order. But I am getting an error in the 4th line which is saying

AttributeError: 'DataFrame' object has no attribute 'date'.

I have 'date' and 'value' attribute in my dataframe. But when I am printing my datacolumns, it is showing only "value" column only. Here is my code:

    y_hat_avg = test_data.copy()
        mod = sm.tsa.statespace.SARIMAX(train_data.value, order=(1, 1, 0), seasonal_order=(1, 0, 0, 12)
                                        , enforce_stationarity=False, enforce_invertibility=False)
        fit4 = mod.fit()
        y_hat_avg['SARIMA'] = fit1.predict(start=test_data.date.iloc[0], end=test_data.date.iloc[-1], dynamic=True)
        plt.figure(figsize=(16, 8))
        plt.plot(train_data['value'], label='Train Data')
        plt.plot(test_data['value'], label='Test Data')
        plt.plot(y_hat_avg['SARIMA'], label='SARIMA')
        plt.legend(loc='best')
        plt.show()

The error is occurring in this line:

    y_hat_avg['SARIMA'] = fit1.predict(start=test_data.date.iloc[0], end=test_data.date.iloc[-1], dynamic=True)

While I am printing only the column, it is showing only value column like this

    print(test_data.columns)
    >Index(['value'], dtype='object')

But if I print the dataframe head, it shows unwanted extra space like this:

    print(df.head())

                    value
    date                
    1996-12-31  0.927377
    1997-12-31  0.927546
    1998-12-31 -0.359870
    1999-12-31  0.537907
    2000-12-31  1.281655

Maybe the error is because of the extra space, but I am not sure.