Your company is operating a home delivery service which requires management of a large fleet of vehicles.
Because of the COVID-related restrictions, demand for your services has significantly increased.
In order to serve this demand with the existing fleet, the company needs to optimize the fleet operation.
You, as a data scientist, are tasked to generate demand forecasts for the next six months.
You decide to use forecasting in the AutoML service.
Which item of the following list is not required when running the forecasting models?
Click on the arrows to vote for the correct answer
A. B. C. D. E.Answer: B.
Option A is incorrect because the most important for time-series forecasting tasks is that training data contains a feature that represents a valid and consistent time series feature having observation data at each point.
Option B is CORRECT because time series forecasting uses Rolling Origin Cross Validation (ROCV) for validating data.
For the ROCV method the training and validation data must be passed together, and the number of cross validation folds must be set.
No separate validation data must be provided.
Option C is incorrect because for time-series forecasting tasks the training dataset must be set in ascending order based on the time feature, so that it represents a valid time series.
Option D is incorrect because it has to be defined for the model how many periods forward you want to forecast.
The horizon must be in units of the time series frequency (days, weeks, months etc.)
It is required and its default value is 1.
Option E is incorrect because the name of the time column is required to specify the datetime column in the input data, which is used for building the time series.
Reference:
When using forecasting in the Azure AutoML service, there are several requirements that must be met to ensure accurate and effective forecasting. The list of items provided in the question includes:
A. Data must contain a time feature for each observation B. Provide validation data for the training C. Data must be sorted in ascending order of the time feature D. Set the Forecast horizon E. Set the Name of the time column.
Let's go through each of these items to determine which one is not required when running the forecasting models:
A. Data must contain a time feature for each observation: This is a necessary requirement for time-series forecasting. Time is a crucial component in forecasting as it allows the model to identify patterns and trends in the data over time. The time feature can be in any format, such as a timestamp, date, or any other time-related feature that can be used to order the data. Therefore, this requirement is necessary for time-series forecasting, and it is not the item that is not required when running the forecasting models.
B. Provide validation data for the training: Validation data is used to evaluate the model's performance during the training process. By providing validation data, the model can assess its accuracy, identify potential issues, and make adjustments to improve the forecast. Therefore, providing validation data is necessary to train the forecasting model effectively. Therefore, this requirement is necessary for time-series forecasting, and it is not the item that is not required when running the forecasting models.
C. Data must be sorted in ascending order of the time feature: Sorting data in ascending order of the time feature is essential because it helps the forecasting model understand the progression of time and the relationship between past and future data. By sorting data in chronological order, the model can identify patterns and trends over time that can be used to forecast future values. Therefore, this requirement is necessary for time-series forecasting, and it is not the item that is not required when running the forecasting models.
D. Set the Forecast horizon: The forecast horizon is the time period over which the model will predict future values. It is an essential parameter to set when running forecasting models as it determines how far into the future the model should forecast. Without setting the forecast horizon, the model will not know how far into the future it should predict, and the forecasts will be invalid. Therefore, setting the forecast horizon is a necessary requirement for time-series forecasting.
E. Set the Name of the time column: Setting the name of the time column is also a necessary requirement for time-series forecasting. The time column is used to sort the data in chronological order, and it is a necessary component in forecasting models. Without setting the name of the time column, the forecasting model will not know which column contains the time data, and the forecasts will be invalid. Therefore, this requirement is necessary for time-series forecasting, and it is not the item that is not required when running the forecasting models.
Based on the above analysis, it can be concluded that all the items in the list are necessary requirements for running forecasting models in Azure AutoML service. Therefore, there is no item that is not required when running the forecasting models.