1
$\begingroup$

I want to forecast the sales for next whole year. Data is available at a day level and I need to forecast at a month level.It is for a Jewelry brand in India. There is seasonality in the data. Also I want to impute flags for festivals,discount offer etc.The model is quite dependent on these flags, as events occur during different time periods in a year.

Which forecasting technique do you think will work best here? I am already using TSLM,ELM,ARIMA,Naive,Seasonal Naive. TSLM & ELM together is giving me avg accuracy of about 10% with a range of 0-25%. Arima/Arimax/Naive are failing big time.

Kindly suggest the appropriate model to be used.

$\endgroup$

closed as too broad by kjetil b halvorsen, mkt, mdewey, Peter Flom Feb 7 at 11:17

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • $\begingroup$ Hi Prachi! Your question is too general. For forecasting, you can start from the Prophet github.com/facebook/prophet $\endgroup$ – wind Feb 7 at 5:43
0
$\begingroup$

This is a very broad question. You likely have in your data (intra-weekly and intra-yearly), though these may be not quite that important if you are only interested in monthly forecasts. Your best bet is likely to use some regression-based approach, using dummies to model the seasonalities and appropriate regressors for offers and events.

This article of mine (Kolassa, 2016, IJF) on retail forecasting, or this review paper (Fildes, Ma & Kolassa, 2019) may give you some ideas.

$\endgroup$
0
$\begingroup$

It probably won't be easy to use all your input variables, such as events etc. I suggest doing the predictions on a day level to have a set of samples for both train and test set. If possible you can also consider to split the sales for different products to get even more rows to your dataset.

Start from linear regression or Prophet, and when you get that to work consider moving to more advanced models.

$\endgroup$
  • $\begingroup$ Hi, I tried Time series linear model (TSLM)- is that same as linear regression for time series data? Any idea? $\endgroup$ – Prachi Feb 10 at 17:07

Not the answer you're looking for? Browse other questions tagged or ask your own question.