CODEWITHSUNDEEP
pythonpandaspandas-groupby
BlueSheepToken
BlueSheepToken

Reputation: 6149

compute relative percentage pandas

I need to compute the relative percentage of each category using pandas, I know I need to use groupby using pandas but I am kinda lost.

Input:

ID | stringValue | FloatValue
A  | 'string'    | 2
A  | 'string2'   | 8
B  | 'string'    | 5

Expected Output:

ID | stringValue | FloatValue | Perc
A  | 'string'    | 2          | 20   
A  | 'string2'   | 8          | 80
B  | 'string'    | 5          | 100

The expected output groups value by their ID and calculate the percrentage. Here, in A you have two value 2 and 8. So the percentage should be 100 * 2 / (2+8) and 100 * 8 / (2+8). For the id B, there is only one value so the Perc should be 100

Upvotes: 0

Views: 593

Answers (3)

Nathan Quinteiro
Nathan Quinteiro

Reputation: 11

Considering your data is a pandas DataFrame named "data", the following code should do the trick:

data["Perc"] = data.apply(lambda x: x["FloatValue"] * 100 / data.groupby(["ID"]).sum()["FloatValue"][x["ID"]], axis=1)

It groups your items by ID and compute the total sum of FloatValue. The apply method of DataFrame then create a new Series by dividing the FloatValue by the corresponding group sum.

Upvotes: 1

Abhi
Abhi

Reputation: 4233

IIUC try:

df['Perc'] = df.groupby('ID')['FloatValue'].transform(lambda x: (x/x.sum()) * 100)

# Output


   ID   stringValue FloatValue  Perc
0   A   'string'      2         20
1   A   'string2'     8         80
2   B   'string'      5        100

Upvotes: 1

Alexis Drakopoulos
Alexis Drakopoulos

Reputation: 1145

If the 5 -> 100% is a typo and you literally just mean how can I make my number look like a percentage, you can do that easily.

If you want it to be % literally then you will have to divide by 10, and you can print it by multiplying by 100.

If you want it to be a percentage and you will keep that in mind, then do the opposite and multiply by 10.

You can do this by simply doing:

df["Perc"] = df["FloatValue"] * 10

Upvotes: 0

Related Questions