Why ipolate leads to one missing value

[Alex Mai]

Sorry for the mismatch between the title and the actual content (I am unable to change the title after I posting it)

I use the command of "ipolate" to extrapolate missing observations, but there appear an odd number. The command is: ipolate v1 v2, gen(v12) epolate.

The first number of v12 is -0.8. that is impossible (v12 couldn't be negative). This is part of the data (for illustration reason, data have also been modified). Thank you!

[Nick Cox]

Your pictures are almost unreadable. We do ask that all posters read http://www.statalist.org/forums/help before posting and do provide quite explicit requests not to post images like this: see #12 in that FAQ Advice.

But your question just seems to pivot on the detail that ipolate needs the epolate option to estimate at the ends of data series. Interpolate in, in the strict sense, means interpolation between data points.

[Alex Mai]

Your pictures are almost unreadable. We do ask that all posters read http://www.statalist.org/forums/help before posting and do provide quite explicit requests not to post images like this: see #12 in that FAQ Advice.

But your question just seems to pivot on the detail that ipolate needs the epolate option to estimate at the ends of data series. Interpolate in, in the strict sense, means interpolation between data points.

Sorry, but I am just now aware about my error in manipulation. What I want to do is to extrapolate. And now I use "ipolate y x, gen(y1) epolate". But I do not know why there appears an odd value (as what i show in the modified post). Thank you very much!!

[Nick Cox]

Alex: I've already reminded you of the FAQ Advice and I do so again.

Data examples should (please) be provided using dataex (SSC).

Code should (please) be presented using CODE delimiters.

If you don't provide readable, intelligible questions, your time too is spent badly as either people struggle to understand you and good answers are delayed or they will just walk away from a poorly posed question.

What does the data listing mean, as there are two columns for some observations and three for others?

But the principle is easy. ipolate interpolates and extrapolates linearly. That's its job. It can know nothing about what is possible or impossible or plausible or implausible: you told it nothing about any of those, nor is there scope to do so.

If you plot your extreme data points:

line v12 v2 in 1/5 || scatter v1 v2 in 1/5

you should see why v12 takes on the value it does in observation 1.

You may need to consider other interpolation methods. Search this forum for mentions of mipolate (SSC).

[Steve Samuels]

When you do want to post a graph, use graph export to create a "png" file and upload that.