Interpolation
Categories: Metrics
Interpolation is a bit like a Mad Libs, but with numbers left out instead of words. It’s the process of using data that already exists to fill in gaps somewhere between the beginning and ending data points where one or more measurements wasn’t taken.
Let’s say that a charity tracked donations for the first three weeks in August and the last three weeks in September, but for some reason they missed tracking their donations in the last week of August and the first week of September. We can try to find the missing values using a bunch of different kinds of interpolation like linear interpolation (the most common method), polynomial interpolation, exponential interpolation, etc.
It might be something as simple as connecting the two data points on either side of our gap with a straight line and locating the data points along that line at the appropriate points. Be careful though: as with Mad Libs, things can go sideways quite quickly. We shouldn’t go all in with our interpolated value as gospel, as it could very well be wrong.