Cubic interpolation

[Carlo Lapaglia]

Hi, it is the first time i post on this forum. I dont know if this is the right place where i can post, but...
I need to interpolate some quarterly data into weekly, can someone help me doing it or tell me where i can find the correct "do file"?

[Nick Cox]

Quarterly data into weekly??? Anything but just replicating the values for as many weeks as there are weeks in a quarter seems hard to defend.

More constructively, see denton from SSC just in case it is what you want.

P.S. We strongly prefer full real names on Statalist, so please use Carlo and your family name.

P.P.S. There are programs cipolate and csipolate on SSC. As their author, I couldn't endorse this proposed application with enthusiasm.

[Rich Goldstein]

linear interpolation can be done with the -ipolate- command; I'm not sure what you mean by "cubic interpolation" which is not explained in your text at all

[ben earnhart]

-help ipolate- will probably get you approximately there. Is your data long or wide?

[Carlo Lapaglia]

I'am sorry i didnt know it was imporatant my last name, but i can't change my user name, i don't know why.

I have quarterly data and i need the weekly data to run a "do file" to calculate var and covar. I think i have wide data from balance sheets of insurance corporates. They told me the most suitbale way to interpolate this data to weekly is to use the cubic interpolation. i dont actually know why cubic interpolation, but i think is different from the linear one.

So do u suggest to simply replicate the missing value orto ipolate usin the linear interpolation?

[Stephen Jenkins]

I'am sorry i didnt know it was imporatant my last name, but i can't change my user name, i don't know why.

Carlo: you can change your user name easily. Hit the "Contact Us" text on the right hand side of the blue bar at the very bottom of your screen, and make your request. The administrators typically deal with requests quickly.

[Nick Cox]

It's your call, whatever you do.

The question is whether you have enough information to do what you want to do well. Interpolating 13 or so values from 1 is quite a stretch. But replicating a quarterly value 13 times is also interpolation. So, you could just try replication, linear interpolation, cubic interpolation, cubic spline interpolation, etc. and use your judgement on what works best.

Changing your identifier is explained in the FAQ Advice, Section 6.

[Carlo Lapaglia]

It's your call, whatever you do.

The question is whether you have enough information to do what you want to do well. Interpolating 13 or so values from 1 is quite a stretch. But replicating a quarterly value 13 times is also interpolation. So, you could just try replication, linear interpolation, cubic interpolation, cubic spline interpolation, etc. and use your judgement on what works best.

Changing your identifier is explained in the FAQ Advice, Section 6.

I've done the "changing-name" request, thanks. Could u post me (or show me) an example of do file for interpolation or replication...i'm really noob at using stata.

[Nick Cox]

No; I won't do that.

People have already mentioned ipolate, which is part of official Stata, and cipolate and csipolate, which can be downloaded from SSC using the ssc command. Each command has a help file which gives examples. The help for csipolate indicates examples which you can just modify for cipolate.

So, you are asking for information that is already available.

[Carlo Lapaglia]

OK thx..i ll look for it...