Get Root mean squared error (RMSE) for a period

Question

I have two time series (forecast and realization) and both start as of 2003 until today. I need to calculate the RMSE for each month and for each year. Could you please help me with the code. I used np.sqrt(metrics.mean_squared_error(actual, forecast)) but it gives RMSE for the whole set. Many thanks!

The data loos like this

I think you need to look into slicing – Shan S Jul 01 '20 at 17:01 — Shan S, Jul 01 '20 at 17:01

score 0 · Answer 1 · answered Jul 01 '20 at 17:03

0

I assume your data is a pandas dataframe. If so, you want to group by months and then aggregate by rmse:

from sklearn.metrics import mean_squared_error
df["differ"] = df["Forecast"] - df["Actual"]
rmseser = df.groupby(['Year', 'Month']).differ.agg(lambda x: sqrt(np.sum(x * x))

answered Jul 01 '20 at 17:03

Igor Rivin

4,632
2
23
35

it seems that it's not working at my end - I got "'DataFrameGroupBy' object has no attribute 'differ'" – Tochiza Jul 01 '20 at 19:59

Get Root mean squared error (RMSE) for a period

1 Answers1