When setting up your Sales Forecasting Configuration, it’s important to understand what each option means. Weighted Averages are simply a way of calculating that gives more weight (importance) to certain elements of the equation. TimeForge offers very customizable Sales Forecasting, and this lesson is designed to help you navigate through set up painlessly. After configuring Sales Categories, you’ll likely want to consider setting up ShiftBuilder Rules so you can start generating schedules with the ShiftBuilder based on your Sales Projections.
Of course, if you need further assistance setting up a forecasting configuration, please don’t hesitate to contact us.
Understanding weighted averages.
When you select Weighted Averages from the Forecasting Configuration menu, you’ll be indicating that you want to customize TimeForge to calculate sales forecasts by your rules instead of our own algorithm. Weighted averages are used in all sorts of calculations, but the place we can all relate to is calculating grades. When you calculate your grade for a class, tests are more important (have a higher “weight”) than quizzes – hence “weighted averages”. When calculating your sales forecast, we can give different weights to each piece of data we consider when forecasting. For example, if you think that last year’s sales are the key indicator of this year’s sales, you’ll want to give a lot of weight to last year’s Actual Sales when you’re setting up your Weight Averages Options.
The following is a breakdown of each option related to “Weighted Averages”.
Base calculations on a day of the week or specific date.
When forecasting for a given day, determine whether the algorithm will find its “base” date (to base its calculations on) by searching for a similar date or a similar day. For example, if it’s forecasting for June 2, 2013, should it look at June 2, 2012 (date) or should it look at the 23rd Sunday of the year?
2. How far into the future should dates be considered?
Determine how many days “forward” from the base date the algorithm should consider. Should it look for these days by date or by day? For the example above, the algorithm will consider the next 7 days (1 week) when making it’s predictions.
3. How far into the past should dates be considered?
Determine how many days “prior” from the base date the algorithm should consider. Should it look for these days by date or by day? For the example above, let’s say the “base date” (determined in option 1) is a Sunday. The algorithm will now consider the previous 7 Sundays when making it’s predictions.
4. How many days prior to the forecasted date should the algorithm factor?
How many days “behind” the day it’s forecasting should the algorithm take into account? Should it look for these days by day or by week? For the above example, let’s say we’re forecasting for today. The algorithm will now consider the previous 7 days when forecasting today. (If I chose “week” it would look at the previous 7 “Sundays” or whatever day of the week the forecasted day is)
5. Enter the weight associated with the next forecasted days.
How important is the number in Step 2? This + step 6 + step 7 + step 8 need to equal 100.
6. Enter the weight associated with the previous forecasted days.
How important is the number in Step 3? This + step 5 + step 7 + step 8 need to equal 100.
7. Enter the weight associated with the days previous to the forecasted date.
How important is the number in Step 4? This + step 5 + step 6 + step 8 need to equal 100.