PlanIQ - New product introduction: all you ever wondered about starting your forecast from scratch

AnaplanOEG

New Product Introduction (NPI) or Cold Start in time series forecasting is a scenario where forecasts should be produced for new items (e.g. new products, new SKUs) for which there is limited historical data, or no historical data at all.

Items with no historical data

In scenarios where the new items have no historical data, it is important to distinguish between two cases:

1. The new items are replacing existing items (manual method):
   If the new items serve as a replacement of existing items, one can use the historical data of the existing items as the historical data upon which the new items’ forecasts will be based. This could be achieved by manually copying the historical data of the existing items to serve as the historical data of the new items. Related data and attributes (static, non-time related categorical data that describes the items) that were connected to the existing item should be copied to the new items as well.
2. The new items are not replacing existing items (automated method):
   If the new items are brand new, use the method involving attributes data and the Amazon Ensemble algorithm. This method requires the use of a preexisting forecast model configured with following settings:
   1. Data collection containing an attributes module and at least 300 historical values (observations) across all time series in the historical data module.
      
   2. Algorithm – Amazon Ensemble.
      The Amazon Ensemble algorithm supported by PlanIQ can generate predictions for new items by leveraging attributes. The algorithm “ties together” items with shared attributes, which in turn allows it to leverage similar items’ historical data to produce forecasts. For example, a clothing company introduces a brand new pair of blue jeans, and the historical data of other jeans or items with a blue color can support the forecast. To use this method, please follow the steps below.
      
      1. Identify an existing item or items that share similarities to the new items in terms of their expected demand, seasonal patterns and/or overall behavior.
         
      2. Add the new items to the item ID list.
         
      3. Populate the attributes module with data for the new items. Make sure that the attributes match as closely as possible to attributes of the item/s you have identified in step a. The engine will use those attributes to make inferences about the new items. In our example above, the model would not be able to predict demand for blue jeans if there are no other items with equivalent attributes.
         
      4. Run a forecast action based on the preexisting forecast model. The results will include forecasts for the new items.

Items with limited historical data

In order to produce forecasts for new items with limited historical data, use either Amazon Ensemble, CNN QR or Deep AR Plus as the forecast model algorithm, as well as relevant attributes data as described above. Alternatively, other algorithms could be used – MVLR, ETS, ARIMA or Prophet. However, these algorithms will not be able to leverage metadata attributes and benefit from the historical data of other similar items.

Once the forecast model has been configured, follow the steps below.

1. Add the new items to the item ID list.
2. If Amazon Ensemble, CNN QR or Deep AR Plus were used, populate the attributes module with data for the new items. The attributes must share some characteristics with preexisting items, as the engine will use those attributes to make inferences about the new items. If other algorithms were used, proceed to the next step.
3. Import the historical data (actuals) of the limited-history items into the historical data module.
4. Run a forecast action based on the forecast model. The results will include forecasts for the new items.

Note that as an alternative approach, one could use a method similar to that described as the manual method under the ‘items with no historical data’ section, and use data of existing items to supplement the data of new items with limited historical data.

Example:

The image of the historical data module below shows the last four periods in the time series data. The column ‘Units_Sold’ holds the historical values. The items Blue Shirt, Orange Sweater and Green Jeans are established items present across all time periods, while the items Blue Jeans and Orange Jeans are “cold start” items. Please consider the following:

• Blue Jeans is a new item with limited historical data. Note that it has non-zero values only for the last two periods.
• Orange Jeans is a new item with no historical data, and has zero values for all periods.

The image of the attributes module below shows that the two new items share attributes with the three already-existing items, and therefore PlanIQ would be able to generate a forecast for those items.

Got feedback on this content? Let us know in the thread below.

Contributing authors: Nitzan Paz, Christophe Keomanivong, Frankie Wolf, Timothy Brennan, Andrew Martin, and Evgenya Kontorovich.