Simple Exponential Smoothing Function and In-Sample Forecasting in NumXL скачать в хорошем качестве

Simple Exponential Smoothing Function and In-Sample Forecasting in NumXL

In NumXL 1.65 (Hammock), you can calculate a fitted or in-sample forecast for your Simple Exponential Smoothing function. Simple Exponential Smoothing Reference: https://support.numxl.com/hc/en-us/ar... For more information, visit us at https://numxl.com/numxl-pro/   / numxl     / numxl   https://twitter.com/intent/follow?sou... _________________________________________________________________ Scene 2: Intro Hello and welcome to the exponential smoothing tutorial series. In our last few tutorials, we discussed how to construct one or multiple steps out of sample forecast and the calibration process for the smoothing parameters for simple exponential smoothing. Now, we will demonstrate how to calculate a fitted or in-sample forecast. But first, let’s discuss what is an in-sample forecast is and how is it different from out-of-sample forecast? The in-sample forecast refers to forecasting an observation that was part of the data sample used to calibrate the model, so it is not really a forecast, more like a model fitted value. For the sample data, we will continue using the sales data of a hypothetical company from the last 2 years. Scene 5: First select the E9 cell, and type in the simple exponential function name: SESMTH(. Scene 6: Once you find the function, click on the “fx” button found on the left of the equation toolbar. This will invoke the function arguments dialog box for the simple exponential smoothing function. Scene 9 For the input data, select the whole cells range corresponding to the sales data which is C9C33 Scene 10 Lock the cells range by pressing F4 Scene 11 For the time order in the input data, type in True or one (1) to designate the 1st observation (C9) as the earliest observation. Scene 12 For the alpha value, select the cell in D1. The value in D1 was calculated during the calibration using the 1st year subset as a training set. Scene 14 Lock the cells reference by pressing F4 Scene 16 And leave the optimization switch disabled by referencing D2. Scene 17 Lock this reference by pressing F4 Scene 27 For forecast time, type in or select a cell with a value equal to the number of steps past the end of the data to include. For Return Type, type in Two (2) for one-step (in-sample) forecast series. Then click OK now. Scene 31-32 The function returns the first value in the array. To display the whole array, select all the cells below then: Press F2 to edit Press CTRL+SHIFT_ENTER Scene 34 The selected cells are now populated with the array values, and the formula is italicized. Scene 38-39 Now, let’s plot the in-sample forecast against our data. Scene 40 And compute the different forecast performance measures. Scene 47 As we can see, in-sample forecast performs worse than out-of-sample. But this is OK, as we would sacrifice in-sample forecast accuracy for good generalization and a better out-sample forecast. Finale That’s all for now. Thank you for watching!