Hello, I have a question about

statsforecast

. Does it take missing values into consideration? From what I have seen so far when testing it, it only seems to use the timestamps of entries for ordering them and not for anything else. At least the cross-validation and the predict method behave this way, I am not sure about the fit method. The cross-validation method seems to iterate over windows of constant size, regardless of how they are placed in time. And the forecast method just forecasts the next K days, regardless of the datetimes specified in the

X_df

argument.

Hi! Is there a method by which we can add events such as holidays to statsforecast? at this moment the holidays are added as columns in my dataframe.

Hi! Quick question: We have seen that Croston, IMAPA, ADIDA, Croston optimized - usually do not change their predictions - no matter how many external variables we add / lagged variables we add. Is this something consistent ? I have read the theory but it seems counter intuitive to me - I mean, they shall change a bit

Hello all! I ve got 2 questions for you guys: 1. Why in mlforecast, if you want to use the

crossvalidation

method from the

MLForecast

class it is done using Time Series cross-validation & in statsforecast, the

crossvalidation

method from StatsForecast class it is done using Expanding Window? 2. I've walked over your tutorial for

crossvalidation

in

statsforecast

(https://nixtla.github.io/statsforecast/examples/crossvalidation.html) and when you want to evaluate the results using RMSE from

datasetsforecast.losses

, I've seen that it is computed on the whole

crossvalidation_df

not on each fold separately & then aggregate the results ( unlike in mlforecast: https://nixtla.github.io/mlforecast/docs/end_to_end_walkthrough.html, where you've put an example on how to compute the losses on each fold separately using

evaluate_cf(df):

function

Hey all, I have a couple of simple questions 1. Why neuralforecast, statsforecast, mlforecast are different libraries? Are models reusable between the libraries? If I wanted to use a neural network from neuralforecast inside my statsforecast pipeline, will I be able to?

2. uber/orbit can receive arbitrary sparse training data (e.g.

train_ds=[2010, 2011, 2014, 2016]

) I was trying to build an adapter for orbit to be integrated into the nixtla model ecosystem (implement custom adapter class with

fit, predict

). However, I can’t seem to find examples where

train_ds

is also accessible, not just

train_y

. It seems that the only other option would be to (1) fill in the missing values in

y

with NaNs, so that the time series would be uniformly spaced with given freq, say

freq='Y'

, or (2) enforce myself to only uniformly spaced data in nixtla libraries (why isn’t this enforced already? Am I missing something)

Hello everyone, I am using statsforecast library fit and predict methods on the dataset. It is throwing an error.

It says that the data is Nonseasonal data

New paper from Smyl where he claims new variant of ES_RNN has outperformed NBEATs. https://arxiv.org/pdf/2212.09030.pdf Code -> https://github.com/slaweks17/ES-adRNN-with-Context

Hello everyone! I've got a question for you guys. If we want to use

differences=[]

do we need to do some scaling on data?

Hello! We’re working on a couple of lightweight open sources TS libraries and we’re wrapping StatsForecast’s models - we love it so far! On another note, does anyone have any idea why AutoARIMA would be repeating exactly the same value when .predict() is called, no matter the forecasting time horizon? (with any of the data we’ve passed in so far) I’ll be working on a script that reproduces the issue, but wanted to ask if I’m making a fundamental mistake somewhere? The other models seem to be working.

Hi all, we would like your support here: https://www.linkedin.com/posts/activity-7012165119644180480-ewnO?utm_source=share&utm_medium=member_desktop.

My personal vote is for Nixtla 🚀 will be interesting to see final results

Awesome Time Series has started to have Python section filled up.

Thank a lot Valeriy!!!

Stumbled upon this one https://github.com/alsabtay/ATAforecasting claims performed well on M3 and M4. Claims superior to ETS and ARIMA

It's already done ! 👍

I'm interested in including exogenous variables using AutoETS and how this was implemented. When I looked at the source code, it seemed like X is not used during e.g. 'fit()' or 'predict()'. Should I conclude from this that AutoETS cannot use exogenous variables? Is there an overview that indicates which models can('t) handle exogenous variables?

It is possible to find a list of which models are able to handle exogenous variables for the StatsForecast library?

Do Nixtla libraries have tests for trend and stationary and decomposition under the hood? Was wondering if some preprocessing is done in terms of trend and seasonality and what sorts of statistical tests the libraries use @Max (Nixtla) @fede (nixtla) (they/them) please?

Hi. Quick q: Is there any way to work around this issue ?

py.warnings WARNING /home/andreitulbure/anaconda3/envs/python-translation/lib/python3.10/site-packages/statsforecast/ets.py:1141: UserWarning: I can't handle data with frequency greater than 24. Seasonality will be ignored.

wait does

neuralforecast

accept a tz-aware

ds

column!? it seems to work locally but has anyone tried this?

Hi, are there any resources of using HistoricAverage? I would like to use the 4w/16w /52w simple moving average but I do not seems to knack the implementation. I would want it to be a rolling average.

Hello, why we don't use the cross validation method in exogenous variable tutorial? Just making sure I'm not implementing something incorrectly.

How is it that the SeasonalNaive forecaster takes a 'level' argument in its predict method? I thought seasonal naive just takes the value from the last cycle which it finds by looking at the season_length. How can it calculate confidence intervals? How does 'level' work for it?

For AutoARIMA, if I set stepwise=False, and parallel=True like the documentation suggests, I get an error of NotImplementedError: parallel=True. Is the method not implemented at all or is it the case for my machine and Python version. I am on Mac M1 and Python 3.10.

How do we get the p,q,d from the autoArima model?

Hello all! I have 2 questions for you guys: 1. I guess WindowAverage model from statsforecast is Simple Moving Average, am I right? 2. Do you have a Exponential Moving Average model?