Get 3 rows from a Pandas Dataframe with the latest timestamp entry for a specific condition

I have put the following log data to a pandas dataframe. Now I want to get the 3 rows with target_label= 'Security Alert' having the most recent timestamps

In the following table the third column is the target_label column,

1759,12/29/2025 20:33,ModernCRM,Unauthorised access attempt from 192.168.99.79 detected,Security Alert,bert
2213,9/1/2025 10:30,AnalyticsEngine,Security alert: suspicious activity from 192.168.214.63,Security Alert,bert
168,5/19/2025 7:56,ModernHR,Multiple incorrect login attempts were made by user 7918,Security Alert,bert
1844,6/10/2025 8:39,ThirdPartyAPI,"Alert: server 9 experienced unusual login attempts, security risk",Security Alert,bert
961,4/26/2025 5:21,ThirdPartyAPI,API access audit trail shows unauthorized entry for user 5627,Security Alert,bert
1077,3/4/2025 9:49,AnalyticsEngine,"Anomalous activity identified on server 47, security review recommended",Security Alert,bert
1356,5/3/2025 13:03,ThirdPartyAPI,Multiple rejected login attempts found for user 1805,Security Alert,bert
43,11/22/2025 11:06,BillingSystem,"Abnormal system behavior on server 40, potential security breach",Security Alert,bert
2062,6/7/2025 1:22,AnalyticsEngine,"Server 11 experienced potential security incident, review required",Security Alert,bert
769,4/28/2025 4:07,ModernHR,API access denied due to unauthorized credentials for user 5914,Security Alert,bert

Solution

First get all rows with target_label='Security Alert'

selected = df[ df['target_label']=='Security Alert' ]

Next sort values by date (doc: pandas.DataFrame.sort_values)

selected = selected.sort_values(by="date")  # .sort_values(by="date", ascending=True)

Finally get last (or first) three rows (depends on sorting "ascending" or "descending")

selected = selected[-3:]  # selected[:3]

So it can be

selected = df[ df['target_label'] == 'Security Alert' ].sort_values(by='date')[-3:]

Of course you have to check if pandas keeps date as datetime object - or you will have to convert them on your own (pandas.to_datetime())

Minimal working example.

I use io only to create file like object in memory - so everyone can simply copy and run this example (but you should use filename)

I had to use pandas.to_datetime() to convert string with date to object datetime.

data = '''1760,12/29/2025 20:33,ModernCRM,Unauthorised access attempt from 192.168.99.79 detected,Security Alert,bert
2213,9/1/2025 10:30,AnalyticsEngine,Security alert: suspicious activity from 192.168.214.63,Security Alert,bert
168,5/19/2025 7:56,ModernHR,Multiple incorrect login attempts were made by user 7918,Security Alert,bert
1844,6/10/2025 8:39,ThirdPartyAPI,"Alert: server 9 experienced unusual login attempts, security risk",Security Alert,bert
961,4/26/2025 5:21,ThirdPartyAPI,API access audit trail shows unauthorized entry for user 5627,Security Alert,bert
1077,3/4/2025 9:49,AnalyticsEngine,"Anomalous activity identified on server 47, security review recommended",Security Alert,bert
1356,5/3/2025 13:03,ThirdPartyAPI,Multiple rejected login attempts found for user 1805,Security Alert,bert
43,11/22/2025 11:06,BillingSystem,"Abnormal system behavior on server 40, potential security breach",Security Alert,bert
2062,6/7/2025 1:22,AnalyticsEngine,"Server 11 experienced potential security incident, review required",Security Alert,bert
769,4/28/2025 4:07,ModernHR,API access denied due to unauthorized credentials for user 5914,Security Alert,bert
'''

import pandas as pd
import io

file_like_object = io.StringIO(data)
df = pd.read_csv(file_like_object, names=['number', 'date', 'system', 'description', 'target_label', 'person'])

print('data type for column "date":', df['date'].dtype)

# convert string with date to datetime object
df['date'] = pd.to_datetime(df['date'])   

print('data type for column "date":', df['date'].dtype)

#selected = df[ df['target_label']=='Security Alert' ]
#selected = selected.sort_values(by="date")
#selected = selected[-3:]

selected = df[ df['target_label'] == 'Security Alert' ].sort_values(by='date')[-3:]
print(selected)

Result

data type for column "date": object
data type for column "date": datetime64[ns]

   number                date           system                                        description    target_label person
1    2213 2025-09-01 10:30:00  AnalyticsEngine  Security alert: suspicious activity from 192.1...  Security Alert   bert
7      43 2025-11-22 11:06:00    BillingSystem  Abnormal system behavior on server 40, potenti...  Security Alert   bert
0    1760 2025-12-29 20:33:00        ModernCRM  Unauthorised access attempt from 192.168.99.79...  Security Alert   bert

EDIT:

If you need it as timestamps

# divide by `10**9` to convert `nanoseconds` to `seconds`

timestamps = selected['date'].astype(int) // (10**9)
print(timestamps)

# check if timestamp gives correct datetime (BTW: problem can be timezone)

import datetime

print(datetime.datetime.fromtimestamp( timestamps.iloc[0] ))

See other methods on Stackoverflow: pandas datetime to unix timestamp seconds