Adding column to pandas data frame that provides labels based on condition - pandas

Question

I have a data frame filled with time series temperature data and need to label the equipment status as 'good' or 'bad' based on the temperature. It is 'good' if it is between 35 and 45 and 'bad' otherwise. However, I want to add a condition that if it returns to the appropriate temperature range after being listed as 'bad' it must be 'good' for a least 2 days before it is labeled as 'good' again. So far, I can label on a more basic level, but struggling with implementing the more complicated label switch.
df['status'] = ['bad' if x <35 or x >45 else 'good' for x in df['temp']]
Any help would be greatly appreciated. Thank you.

Answer 1

What about an approach like this?
You can make a group_check function for each row, and check if that row has any neighboring offending temperature within the group from the broader df.
This will only check the previous measurements. You would need to do a quick boolean check against the current measurement to confirm the prior measurements are OK AND the current measurement is OK.
def group_check_maker(index, row):
def group_check(group):
if len(group) > 1:
if index in group.index:
failed_status = False
for index2, row2 in group.drop(index).iterrows():
if (row['Date'] > row2['Date']) and (row['Date'] - row2['Date'] < pd.Timedelta(days = 2)) and (row2['Temperature'] < 35 or row2['Temperature'] >45):
failed_status = True
if failed_status:
return 'Bad'
else:
return 'Good'
return group_check
def row_checker_maker(df):
def row_checker(row):
group_check = group_check_maker(row.name, row)
return df[df['Equipment ID'] == row['Equipment ID']].groupby('Equipment ID').apply(group_check).iloc[0]
return row_checker
row_checker = row_checker_maker(df)
df['Neighboring Day Status'] = df.apply(row_checker, axis = 1)

Answer 2

import numpy as np
df['status'] = np.where((df['temp']>35) | (df['temp']>45) , 'bad', 'good')
This should solve the issue.

Answer 3

you can create a pd.Series with the value 'bad' that you replace values where its is betweem 35 and 45, then propagate the value 'bad' to the next two empty rows with ffill and a limit of 2 and finally fillna the rest with good such as:
#dummy df
df = pd.DataFrame({'temp': [36, 39, 24, 34 ,56, 42, 40, 38, 36, 37, 32, 36, 23]})
df['status'] = pd.Series('bad', index=df.index).where(df.temp.lt(35)|df.temp.gt(45))\
.ffill(limit=2).fillna('good')
print (df)
temp status
0 36 good
1 39 good
2 24 bad
3 34 bad
4 56 bad
5 42 bad #here it is 42 but the previous row is bad so still bad
6 40 bad #here it is 40 but the second previous row is bad so still bad
7 38 good #here it is good then
8 36 good
9 37 good
10 32 bad
11 36 bad
12 23 bad