Using TIM in Excel
TIM Forecasting in Excel supports RTInstantML. This section explains how to go about making a RTInstantML forecast using TIM in Excel. Before starting the steps in this section, a user is expected to have added TIM in Excel into their Excel application and to have their data in the correct format.
In the 'Home' tab, click on the 'Tangent Information Modeler' button to start TIM in Excel. This will open a taskpane within the Excel application. In the taskpane, fill out your credentials (License Key, email and password). Make sure the 'Remember me' checkbox is checked, as this prevents you from having to fill out your credentials every time either the add-in or Excel is restarted. Click 'Log on' to validate your credentials and move on to the add-in's forecasting capabilities.
The add-in automatically extracts the most plausible dataset range and asks the user to confirm or adjust this range. If multiple plausible dataset ranges exist in a single worksheet, the add-in assumes the dataset is contained in the range closest to the beginning of the worksheet (cell A1). If a user is not satisfied with the extracted range, they can manually change it. This is done either by adjusting the range address shown in the relevant text field in the add-in; or by selecting the desired range in the worksheet. When satisfied, the user needs to click 'Confirm' to confirm the range.
After, confirming the desired range, a menu will be shown below the range address field and range selector button, from which the target variable can be selected. By default, the second column of the dataset is expected to contain the target variable. The user can however select any other column to represent the target variable, with the exception of the first column, containing the timestamps.
The predictors present in the dataset are listed in the taskpane, below the target variable menu. Users can select the predictors they want to include in the calculation of the forecast. The predictors that are unchecked at the time of forecasting will be ignored.
Below the list of predictor variables, the user can indicate the number of samples they want to forecast. This is expected to be a positive integer between 0 and 5001.
After indicating the desired forecasting settings, the user needs to click 'Make forecast' to send the RTInstantML request to TIM's engine. A message will tell the user that the forecast is being calculated, and the user will be informed of the progress via a progress bar and percentage. When the calculation of the forecast is finished, a message will tell the user that the forecast can be found at the bottom of the dataset. If the user is still active in the worksheet in which the forecast was requested, the add-in will scroll down to the newly added values. If the user is active in another worksheet, the add-in will not jump to the forecast, as to not disturb the user. Hint: scrolling down to the bottom of a (long) dataset can be done by clicking any cell in the dataset and subsequently typing Ctrl + Down Arrow (for Windows users) or Cmd + Down Arrow (for MacOS users). In case an orange message box appears, there was a warning during forecasting. The requested forecasts will still appear, but acknowledging the warning can improve the quality of future forecasts.