Forecasting

I'm not an expert in sales forecasting. A quick look at your data with my untrained eye reveals no discernible trend or basis for extrapolating.

Do you know what kind of forecasting algorithm you want to use? Wikipedia lists several types of algorithms: https://en.wikipedia.org/wiki/Forecasting without giving a lot of details about those algorithms. Excel has a few built in algorithms and utilities, including regression algorithms (LINEST() for linear regressions or Solver for non-linear regressions), various FORECAST() functions (that seem based on either straight line linear regression or exponential triple smoothing algorithms), and with enough creativity, one can program most other algorithms into the spreadsheet. I would suggest that the very first step in this sort of forecasting/prediction is choosing a suitable algorithm, which is not something Excel can do for you.

I agree with MrShorty. You need to be able to describe your situation enough so that we can understand what the variables might be. What are the factors that affect sales volume? Is this is a seasonal business? Are you trying to forecast in the same granularity as what your data is? You just have a single column for Forecasted Sales--is that a single figure for all of 2018? Your data is weekly but only for 17 weeks. If that is actual data for 2017, it is not enough data for a reliable forecast for all of 2018.

In short, your problem is not an Excel problem, it's a business problem, and you may not have enough data to solve it.

(I am not an expert in forecasting but I took a graduate course in that subject so I have an understanding of it.)

As indicated, this is well outside what I do and know, so I probably should just let 6StringJazzer respond and bow out of the discussion. However, if I understood Wikipedia's description of three of those four algorithms, then seem so simple that I am uncertain if I understood them correctly. If I understood them correctly, I am surprised that you are having difficulty programming them in Excel.

Naive method: It looks like this method assumes that all forecasted values are the same as the last observed value. Assuming that is the value in column BA, then one formula for forecast values beyond that week might be =$BA2 (note the mixed reference). An HLOOKUP() function could probably also be used, though it would be a lot easier if you had real numbers in row 1 rather than those text strings (or a helper row with real numbers -- perhaps the first date of each week or a number of week since start of period kind of number no matter what algorithm you end up choosing, I suspect that this will be a useful step for most algorithms -- get a useful T variable in the spreadsheet). =HLOOKUP(DATE(2017,10,24),$B$1:$BA$10000,row(),true) assuming you choose a lookup value and first row based on dates.

Seasonal naive: Same thing -- assume all future values for a given "season" (how are you going to define season?) are the last observed value for that season. Again, a simple HLOOKUP() is my first thought.

Drift: Assume a straight line forecast that connects the first and last observation. I would probably extract those two data points into a separate table (using HLOOKUP() or simple direct references), then use SLOPE() + INTERCEPT() or LINEST() to get m and b for this straight line, then use those to compute forecasts.

As for the time series methods, you don't indicate which of those several methods you are wanting to use. As I noted above, Excel has the built in functions that use linear regression techniques to do this (see help files for LINEST(), FORECAST(), TREND(), GROWTH(), LOGEST(), etc.). Excel 2016 introduced the FORECAST.ETS() functions that use a specific exponential triple smoothing algorithm if that is more what you have in mind.

As indicated, the first three of those algorithms seem so simple, I am not sure I understood them correctly. You might spend some more time with those, make sure I understood them correctly, and see what you are having trouble understanding about these algorithms. More detail about your desired time series algorithm will probably be needed before tackling your desired time series algorithm.

Here is an example of one I did for fuel sales. There are two years' worth of data. The business is seasonal and I used a quadratic model. This is just to give you an idea what can be done. There is a chart that shows the original data, then it shows the model with the data predicted by the model. You can see that the model fits the data nearly perfectly, which is unusual.

@dluhut: I use thread subscriptions for that sort of thing. Just above the first post in the thread (off towards the right) is a series of dropdown menus, one named "Thread tools". Select this drop down menu, and one of your options should be "subscribe to this thread". In the page following, click on "add suscription" to add this to your subscription folder. To access subscribed threads, go to User CP (Top of page below excelforum banner). There should be a "view all subscriptions button" as well as a "My subscriptions" box with link(s) to your subscription folder(s).