Pandas interpolate - don't just repeat last values

Asked Aug 01 '19 at 18:15

Active Aug 01 '19 at 18:15

Viewed 103 times

I have a pandas dataframe df that looks like:

      HOUR  VALUE
0     1     2
1     2     4
2     3     6
3     4     None
4     5     None
5     6     None

when I do df.interpolate(axis=0), it returns:

      HOUR  VALUE
0     1     2
1     2     4
2     3     6
3     4     6
4     5     6
5     6     6

Is there something I can do to return:

      HOUR  VALUE
0     1     2
1     2     4
2     3     6
3     4     8
4     5     10
5     6     12

? I am using dummy numbers in this example, but with my real numbers, I just want to get roughly numbers that follow a linear trend, is there something that will generate numbers that follow a linear trend, and I can maybe specify things like standard dev?

asked Aug 01 '19 at 18:15

Preethi Vaidyanathan

1,203
1
12
32

may be a linear fit on `VALUE` against `HOUR`? – Quang Hoang Aug 01 '19 at 18:17
Is that a pandas function besides `interpolate`? – Preethi Vaidyanathan Aug 01 '19 at 18:29
1

No, I think you would need to resolve to others, e.g. `np.polyfit`. – Quang Hoang Aug 01 '19 at 18:29
You want to `extrapolate`. Interpolation is used to fill values `between` non-null values. This requires curve fitting. – ALollz Aug 01 '19 at 19:29

Pandas interpolate - don't just repeat last values

0 Answers0