In my research study, i have four annual time series and i I want to convert them to monthly series while retaining the wide fluctuations in the original series. knowing that the curves of these annual series are approximately exponential.
when confronted with severely positively skewed data i always try the logarithm, whose difference is approximately the rate of increase.
- Lee De Cola can you provide me more explanantion please, if you will permit, I'll be very grateful .
I'm confused Chbili Sfia as if you have an annual times series did you also collect monthly data obviously? I thought if you called it an annual time series it would only be collectly annually.
I am doubtful on the reasons why you want to convert annual time series to monthly time series. There are in the literature methods for doing that if you have also a monthly time series that provides the seasonal pattern. Otherwise it is purely interpolation and there are numerical methods for that purpose. Anyway, don't expect good results from that exercise and don't consider that you will get longer time series. The length increase is, of course, purely artificial.
Deborah J Hilton yes, my research subject is related to a commodity price modeling, which is the price of fertilizer, the production and the demand. for the price, I found monthly data, but for the level of production and the level of demand, I found only annual data, I want to convert these annual data to monthly data knowing that these annual data show well an exponential curve. it is for that that I want to convert them. (there is no monthly data for production and demand level in anywere just anuual observations)
Guy Mélard Yes, I understand what you are saying, but I am obliged to do so because my research focuses on modelling the price of fertilizers, taking into account the explanation of the variation in production volume and demand for this raw material. For the price, I found monthly data, but for the volume of production and demand, there are only annual data, that's why I want to convert them.
Yes, I know that these converted data will be artificial and biased, but I'm going to take these errors into account in the explanation.
I see no reason that the monthly production pattern will follow the yearly production pattern. It depends if the production is inside or outside the country. If it is inside, I suppose that the production will be concentrated just before the product will be demanded and since it is a product for the agriculture, it will depend on when it will be needed. If you know the producers, why not asking them? If it is outside the country, it is more difficult.
It is not theoretically possible to create such a series, otherwise, such a series will not be adequate to the action, since in real monthly series there is periodicity.
No Viktor Zadachyn it is possible provided you have a related monthly time series with the proper periodicity. Of course it will differ from the true (unknown) monthly time series.