Assign dictionary key and priority value wherever conditions from a dictionary are met

user_6495910 0
Viewed 43

I have a dictionary of conditions called rules, which I apply to a dataframe df. Using numpy's select(), I create a new column in df with the dictionary keys where ever the first condition is True. The code is as follows:

import pandas as pd
import numpy as np

df = pd.DataFrame({'col1': [1, 2, 1, 3], 'col2': [4, 4, 4, 3]})
rules = {"Alert 1": df["col1"] == 1,
         "Alert 2": df["col2"] == 4}
df['alert'] = np.select(rules.values(), rules.keys(), default = None)

df
Out[2]: 
   col1  col2    alert
0     1     4  Alert 1
1     2     4  Alert 2
2     1     4  Alert 1
3     3     3     None

I would like to change the dictionary rules such that it consists of vectors that contain the original conditions plus a priority value. In addition to the dictionary key being written to df, I would like this priority to be written as well. Modification to rules, as well as my attempt to write both the dictionary key and priority to df:

df = pd.DataFrame({'col1': [1, 2, 1, 3], 'col2': [4, 4, 4, 3]})
rules = {"Alert 1": [df["col1"] == 1, "High"],
         "Alert 2": [df["col2"] == 4, "Medium"]}
df['alert'] = np.select(rules.values()[0], rules.keys(), default = None)
df['priority'] = np.select(rules.values()[0], rules.values()[1], default = None)

I get an error.

Ideally, I would like the output

   col1  col2    alert  priority
0     1     4  Alert 1      High
1     2     4  Alert 2    Medium
2     1     4  Alert 1      High
3     3     3     None      None

Is there a way to accomplish this?

P.S. I need to keep the priority with the condition in the dictionary. I don't want a separate dictionary which maps the priority onto the dictionary key.

pandas select numpy python
1 Answers

If you wanted to stick with your current approach, you could use tuples to contain all of the values you need for each key. In this case you just need to pull the values at index 0 for the alert and map the resulting values to index 1 for the priority

import pandas as pd
import numpy as np

df = pd.DataFrame({'col1': [1, 2, 1, 3], 'col2': [4, 4, 4, 3]})
rules = {"Alert 1": ([df["col1"] == 1, "High"]),
         "Alert 2": ([df["col2"] == 4, "Medium"])}

    
df['alert']  = np.select([x[0] for x in rules.values()], rules.keys(), default = None)
df['priority'] = df['alert'].map({k:v[1] for k,v in rules.items()})

Output

   col1  col2    alert priority
0     1     4  Alert 1     High
1     2     4  Alert 2   Medium
2     1     4  Alert 1     High
3     3     3     None      NaN
user_47183500
Related