Hello. Could someone please help me? I want to use the informer model available in the neural forecast in my own csv file. I have multivariate data in which columns are there in this way: date, var1(t-1), var2(t-1), and var1(t) where var1(t) is the target variable and others are feature variables except the date. I want to split my data into test and train only and do prediction analysis. Plus, I want to save my data in csv file. Someone please provide the code or assistance with it. I'll be very grateful to you.

It would be something like:

import pandas as pd
from neuralforecast import NeuralForecast
from neuralforecast.models import Informer

model = Informer(h=12,
                 input_size=24,
                 futr_exog_list=[var1,var2])

nf = NeuralForecast(
    models=[model],
    freq=your_data_freq
)

nf.fit(df=training_set)
forecasts = nf.predict(
forecasts.to_csv(path)

The documentation for Informer is here.

Thank you so much for your reply. One more thing, I've attached my dataset. Can you please let me know if I have to make some changes as per the code requirement? This is my dataset.

In the code, we do expect the date column to be called

ds

and the target column to be called

y

. Maybe changing those column names will make it easier

Okay. I've done changes in my dataset. But I am getting error. Here is my code import pandas as pd from neuralforecast import NeuralForecast from neuralforecast.models import Informer from neuralforecast.losses.pytorch import MAE from sklearn.model_selection import train_test_split Y_df = pd.read_csv('lai_supervised.csv') Y_df['ds'] = pd.to_datetime(Y_df['ds']) X = Y_df[['var1(t-1)', 'var2(t-1)']] Y = Y_df['y'] # Split data into train and test sets train_X, test_X, train_y, test_y = train_test_split(X, Y, test_size = 0.2, random_state = 0) training_df = pd.DataFrame({ 'ds': train_X.index, # Assuming train_X has a datetime index 'var1(t-1)': train_X['var1(t-1)'], 'var2(t-1)': train_X['var2(t-1)'], 'y': train_y }) model = Informer(h=12, input_size=24, futr_exog_list=[['var1(t-1)','var2(t-1)']] ) nf = NeuralForecast( models=[model], freq=1 ) nf.fit(df=training_df) forecasts = nf.predict() forecasts.to_csv('forecast.csv') error is related to unique_id. Kindly help me

The library expects a

unique_id

column that identifies each series. If you have a single series, the column should have a constant value. Please, read our quickstart guide to get started with neuralforecast.

Thank you for your reply