After analyzing the data...For all 4 quarters in 2022 prepare a forecast of demand (Can use Excel but please show work/formulas)

Also please explain which forecasting model is best and why. Thank you!

Software Packages Demanded for Entire Market When historical demand data are available, there's nothing like a plot of those data for getting started. If there's a pattern to the demand we can easily plot for it. The plots also help us set the initial values for doing the forecasting in a way consistent with the historical data. If, for example, plots reveal seasonal factors, we can estimate the base value by take the average for at least one seasonal cycle. We can find seasonal indexes by averaging the indexes calculated for each period in the cycle. Similarly, a plot of the values for trend data would enable us to draw in a trend line (or we could average the period-to-period changes) to estimate the trend value. The plot would also help determine the base value to for forecasts. In every instance (constant data, trend data, or seasonal data), plots will help us determine whether it's desirable to use the more recent data in setting starting values. Once starting values have been determined, we can use relatively high smoothing constants for the early forecasts to quickly overcome any errors in the starting values. It's also desirable to make simulated forecasts of the past few periods of historical data to estimate initial values, and then simulating forecasts for the remaining 25 percent of the data, values for starting the initial forecasts would already have been smoothed by the forecasting model. The choice of smothing constants for use in the models for forecasting is a matter of balancing responsiveness with stability. This isn't an easy balance, however, and practice has provided some guidance. For smoothing the average base value, an of about .1 to .2 has been found useful in practice. The value is generally held to less than the value, about .05 to .1 . The value for depends on how frequently the seasonal index is recalculated. If often, such as every few weeks, a low (.1) is acceptable. If less frequently (yearly), =.3 to .4 might be used. In practice, some simulation with past data can be useful. However, we feel this approach is of limited value, since the objective is.to forecast well in the future. The issue always comes down to the stability-responsiveness trade-off, based on how stable the future environment is judged to be