Pandas 'min' aggregate, but how to set 'min'+1?

Question

I’m trying to aggregate my pandas dataframe with minimum and also minimum+1. Let me explain.
Suppose I have a dataframe:

  distance  vertex  type
0   8       104     A
1   1       114     A
2   1       103     B
3   2       102     A
4   3       18      A
5   3       108     B

I get the minimum distance with groupby on Type as follows:

mask = df['distance'].isin(df.groupby(['type'])['distance'].agg('min').values)
df[mask]

This gives me minimum distance for each Type.

  distance  vertex  type
1   1       114     A
2   1       103     B

My question: How do I get rows that are satisfying the following formula:

distance = minimum(distance) + 1

This is what I’m trying to get.

  distance  vertex  type
1   1       114     A
2   1       103     B
3   2       102     A

Asked By: Mohd Salman

||

Source

Answer 1

The exact logic is unclear, but assuming you want to filter per group to keep the values equal to the minimum or minimum +1, use groupby.transform:

mask = df['distance'].le(df.groupby(['type'])['distance'].transform('min').add(1))

out = df[mask]

Output:

   distance  vertex type
1         1     114    A
2         1     103    B
3         2     102    A

Answered By: mozway

Answer 2

This is what I wrote at first, but its wrong and here just as a "Dead code" for people to see and be aware :). See comments why its wrong..

mask = df['distance'].isin([df.groupby(['Type'])['distance'].agg('min').values[0], df.groupby(['Type'])['distance'].agg('min').values[0]+1])

Answered By: Mohd Salman

Answer 3

Another possible solution:

m = df.groupby(['type'])['distance'].min().unique()
mask = df['distance'].isin(np.concatenate([m, m + 1]))
df[mask]

Output:

   distance  vertex type
1         1     114    A
2         1     103    B
3         2     102    A

Answered By: PaulS

Pandas 'min' aggregate, but how to set 'min'+1?

Question:

Answers: