I am training an ML model on multiple unique ids, but when i am trying to forecast, i am trying to forecast for one unique_id only because forecast horizon is different for each unique id & there are external regressors data also in it.. hence i tried something like this when generating forecast # Create a dataframe with data for only the ID you want to forecast single_id_df = df[df['unique_id'] == 'specific_id'] # Generate forecasts forecasts = fcst.predict(h= horizon, X_df =single_id_df) However, i am getting following error..

ValueError: Found missing inputs in X_df. It should have one row per id and time for the complete forecasting horizon.
You can get the expected structure by running `MLForecast.make_future_dataframe(h)` or get the missing combinatins in your current `X_df` by running `MLForecast.get_missing_future(h, X_df)`.

I even checked X_df for any missing inputs, it has no missing inputs & length of X_df is same as h.. Why i am getting this error, can anyone pls help..

It expects all ids to be present. Can you try setting

ids=['specific_id']

in the predict call?

tried, it is still giving same error 😞

The following works as expected:

from mlforecast import MLForecast
from sklearn.linear_model import LinearRegression
from utilsforecast.data import generate_series
from utilsforecast.feature_engineering import trend

freq = 'D'
h = 5
data = generate_series(10, freq=freq)
train, future = trend(data, freq=freq, h=h)
fcst = MLForecast(models=[LinearRegression()], freq=freq)
fcst.fit(train, static_features=[])
fcst.predict(h=h, X_df=future, ids=[0])

You must truly have missing combinations for that id

fcst.get_missing_future(h=h, X_df=X1_df) I used this to check for missing data, i got None as answer.. I checked X_df.isna().sum(). Still no null values in X_df..

If you can provide a reproducible example I can help further. The example I provided works

Hi @José Morales, you are correct.. there looks to be issue with that id.. other ids are working..only that id is not working..

Keep in mind that not only the shape matters but the times. If there are missing times for that id you'll get an error, the times have to be consecutive

you are again bang on! There were few missing dates, hence those ids were giving error.. thanks again