Help for beginner - forecasting events

#1
Hi,

I'm looking for general pointers regarding forecasting events.

We have a process where sales of services are made. These sales go through a lengthy admin and approval process.

I'm looking to predict how many approved sales will complete the process in the current month.

I have a table of data showing the number of Sales each day for the previous 2 years. I also have data for approved sales by day for the previous year.

The processing by admin is seasonal i.e. during the month the tickets peak in week3-4. Additionally there is some seasonality in months i.e. in December less admin resource is available

The primary variable determining the number of approved sales is the sales pipeline, which changes everyday.

Other variables are: the number of admin people available, quality of the sales application form, 3rd pary admin resource

Normally, I would just predict the numbers using rudimentary maths such as Approved Sales = Lagged Sales * Avg Approval %
This is like saying, ignoring all other things the maximum approved sales should be x.

I've created a chart and can in early months there was a linear trend. More recently this trend has flattened out somehwhat although a small trend is still present.

Now I want to improve my forecast.

I don't have complete historical data for all of the the variables, particularly the 3rd party resource levels and quality of app is a more qualiative variable.

Under this case what type of statistical approach would you suggest?


At first I looked at calculating seasonality but I ran into problems because some months have 4 weeks and some have 5 additionally, weeks don't fit months well.

I apprecicate this is a wide question. I want to complete this work to improve my skills and will be happy to share my results here.
 
Last edited: