In fact, above this is holt-winters no seasonal trend model,the above s (t) corresponds to the following a (t)-intercept (smoothed value)B (t) still corresponds to B (t)--trend,t corresponds to K. Alpha corresponds to AlphaGamma corresponds to Beta tower
since (t)-hat is a function of alpha and gamma , TSS is a function of alpha and gamma. to make it easy to understand and manipulate, we use Excel to solve it. The original data is as follows:The results of the design table are as follows:which means we set the initial value, s1=143,b1=-0.65think of H2 and H3 as cells whose values change, and H4 are the cells that we want to get the minimum value by changing the values of H2 and H3. Enter in C3: = $H $2*b3+ (1-$H $) * (C2+D2)corresponds to the formula (1)Enter in D3: = $H $3* (C3-C2) + (1-$H $) *d2corresponds to the formula (2)the remaining values of the C and D columns are then automatically populated. Enter in E4: =c3+d3corresponds to the formula (3), then the equivalent of T all take 1, with the previous period forecast under expectations. Enter in F4: = (b4-e4) ^2the formula corresponding to TSSwhy the initial value of the other t+t is 3, because T constant is 1, then t=2,3,4 ...then the initial value of the t-1 is 1, otherwise (1) (2) (3) The bidding clubs of the formula appears 0 at this time because the values in H2 and H3 are not determined, so the above calculation results are problematic, do not panic. Here we use the Solver feature of Excel:If Solver is not found under the data column, then the file--options--add-ins--excel add-ins--go to-tick solver-click OK. but! If you're using a office64 bit, the problem is:Unable to run the SOLVER. xlam! Getsolverlabel "macro. This may be because the macro is not in this workbook ...I spent a morning over the various forums and Baidu know, have not found the answer, finally, to test the next classmate's office2013, found his can, copy the corresponding file is invalid, finally found that it uses 32 office, so changed a 32-bit office2016, Then solve this problem, well, the permission to the macro also open it. Click Solver and follow the inputs. we need to enter the target location, even the cell with the lowest value of TSS. The input variable cells are H2 and H3 so that their values change and eventually get the minimum value of TSSClick Add to add a constraint. Finally, click Solve. If you are using it for the first time, then it is likely to pop upthe file did not find the Solver32.dll error. DLL should be a bootstrap file. Surprisingly , I used the installation version, after double-click Run, are installed immediately, do not give me the opportunity to choose (so sad in the C drive), after loading, but also run the bat file, the retail version of the VL, and eventually run the corresponding Kms.cmd file. My boot file is in the
C:\Program Files (x86) \microsoft Office\root\office16\library\solverThis problem can be resolved by copying to C:\windows. Finally, the results are:Optimal Smoothing Parameter smoothing parameter α=0.38,γ=0.20, at this time the error and 5870.28 Reference Documents:The application of Excel in the optimization of the parameters of Holt exponential smoothing _ Wu XiaomingMatching of
from for Notes (Wiz)
List of attachments
Solving holt-winters parameters using Excel in conjunction with linear programming
