Dear Alian,
There are many methods for sales forecasting. Some examples include:
1. Delphi methid
2. Time series analysis (Moving average, Exponential smooting, Trend projection)
3. Neural Network
4. Data Mining
5. Conjoint analysis
6. Fuzzy Neural Network etc.
You can find numeorus papers on each topic. Now select your topic of interest and go through some papers to have ideas on the topic.
Hope this helps you.
Regards,
Ali
Dear Asthma,
for sale forecasting it is often a mix of models, for example during exceptional periods such national holidays or 1st January you are in one outlier situation and some models will not adapt well for those period. As it is one exceptional period the data is limited and then you can use time series with autoregression for example or GAM (I am bias here as I find this modelling method very versatile).
For recurring period, such a normal day of week you can use time series or for example boosting for this slice of period and you reproduce for all the periods, for example one for monday, tuesday etc... .
Hope this could help.
Alain
Thank you very much for such useful help I appreciate it .
Basically, I am thinking of using linear regression analysis to predict the future sales, I will use R statistical tool .Do you agree with me and think it is a good idea ?
I doubt linear regression will suffice but I am no expert in sales forecasting. It depends on your particular domain (e.g., actors, commodities, markets etc).
It could for short time forecasting but as Jonas mentioned it is domain dependant. I would be more tempted to use non linear methods, but for example with repetitive periods I managed to do quite good forecasting with linear model (sometime non parametric) and few variables. One advantage of lm in R the speed and if you want to explore before to build more complex models it is a big plus.
- Badges
- Science topic
- Similar topics
- Cell Biology
- Methods