We recommend you use this information to help you select an algorithm that best suits your data collection and the type of predictions you want to generate.
PlanIQ supports the following algorithms:
Anaplan AutoML chooses the best algorithm for your forecast, based on the properties in your dataset. Anaplan AutoML applies all types of Amazon Forecast algorithms. This includes the different variants of DeepAR+ with different hyper parameter configurations.
PlanIQ compares MASE (mean absolute scaled error) results for every individual item for every algorithm that is used. It then chooses the algorithm that provides the best MASE result for the majority of the items. This algorithm is used for the forecasts.
To summarize, Anaplan AutoML chooses the algorithm that is the best for the majority of items in the data collection. This is rather than for every item in the data collection.
For example, in a data collection with 10 items, Anaplan AutoML picks the algorithm that provides the best MASE result for the majority of items.
Based on MASE, Algorithm A provides the best forecast for three items, and Algorithm B provides the best forecast for eight items. The engine chooses Algorithm B as it provides a better performance for the majority of the items.
Amazon Forecast AutoML
Amazon Forecast checks all of the Amazon algorithms and applies the most suitable one for your data collection. Amazon Forecast AutoML selects an algorithm that provides the best prediction performance for the entire data collection based on an assessment of the performance at P10, P50, and P90 quantiles.
Both Anaplan AutoML and Amazon AutoML choose the single best performing algorithm for the entire data collection, but the optimization and selection criteria is different.
This is a recurrent neural networks (RNN) model for larger data sets that contain hundreds of time series. It works with related data that has forward-looking information and attributes.
This is a Convolutional Neural Network - Quantile Regression model for larger datasets that contain hundreds of time series. It's the only algorithm that accepts related data with or without forward-looking information. The algorithm also accepts attributes.
This is an Exponential Smoothing algorithm. ETS is useful when simple concepts of trend and seasonality are likely to explain most of the variance in the time series data. Since this algorithm processes each list item separately, it's unable to take advantage of related data or attributes.
Note: We recommend you use a different algorithm for datasets that have a weekly time scale and yearly seasonal patterns.
This is an Autoregressive Integrated Moving Average algorithm. It's a statistical model useful for datasets that can be mapped to stationary time series. ARIMA can be used when simple concepts of trend and seasonality are likely to explain most of the variance in the time series data. Since this algorithm processes each list item separately, it's unable to take advantage of related data or attributes.
This is an additive regression model. Use Prophet for scenarios that involve seasonality and outliers. It works with related data that has forward-looking information.
CNN-QR and DeepAR+ are global algorithms. They look not only at individual time series in order to predict, but also learn from other time series from the same dataset. This is particularly useful in cases where multiple time series share similar behaviors.
Here is a summary of the technical differences between the individual algorithms:
|Able to use related time series||Yes||Yes||Yes||-||-|
|Able to use related time series without forward looking information||Yes||-||-||-||-|
|Able to use item metadata||Yes||Yes||-||-||-|
|Suitable for what-If analysis||Yes||Yes||Yes||-||-|
|Suitable for cold-start scenarios||Yes||Yes||-||-||-|
|The forecast model is updated with every prediction||-||-||Yes||Yes||Yes|
|The forecast model needs to retrain to incorporate new actuals into the model itself||Yes||Yes||-||-||-|