Interpret Mean vs. SMAPE results [duplicate]

Question

I have following dataframe:

  Impute Technique   Mean    Std   SMAPE
1   EINSTEIGER_ROH   0.00   0.00     NaN
6             MEAN  20.18   8.83  123.40
5      KNN_IMPUTER  20.30   8.79  126.56
7           MEDIAN  22.12   9.30  112.35
3      INTERPOLATE  28.91  10.56  120.24
2            FFILL  35.67  11.27  112.86
0            BFILL  41.01  22.54  112.98

I tried different Imputing techniques (for missing values) and wanted to evaluate which one worked best on average. In column "Impute Technique" you can see the different methods, and column "Mean" represents the Mean Average Error, "Std" the mean standard deviation and "SMAPE" the results from a SMAPE calculation (I couldn't use MAPE because of 0 values in y_pred and y_true).

Now I can't really explain why e.g. for the "Mean" column the "MEAN" method worked best but for the "SMAPE" column the "MEDIAN" method worked best. How can this be explained?

For calculating the SMAPE, I used following function:

def smape(A, F):
    return 100/len(A) * np.sum(2 * np.abs(F - A) / (np.abs(A) + np.abs(F)))

Thanks and regards!