I have huge time series dataset containing TIME, POWER as columns, I'm taking this dataset in window of 30 minutes and inside each window if POWER value is 0 for greater than 10 minutes continuously, then I'll store the start_time, end_time, duration, and number of occurrences of such instance within the window in the dictionary.
window_size = 30 # Identify anomalous instances within the window anomalous_ranges = [] for i in range(0, len(df) - 1, window_size): batch = df.iloc[i:i+window_size] # print(batch) start_time = None end_time = None for idx, row in batch.iterrows(): p = row['ACTIVEPOWER'] if p == 0: if start_time is None: start_time = row['TIME'] else: end_time = row['TIME'] else: if start_time and end_time is not None: duration = (pd.to_datetime(end_time) - pd.to_datetime(start_time)).total_seconds() / 60.0 print(duration) label = 'Zero Values' if duration >= 10: anomalous_ranges.append((start_time, end_time, duration,label)) start_time = NoneOutput:anomalous_ranges
Start Time End Time Duration (Minutes) Remarks0 2023-04-19 01:37:30 2023-04-19 01:53:30 16.0 Zero Values1 2023-04-19 14:37:30 2023-04-19 14:50:30 13.0 Zero Values2 2023-04-19 15:37:30 2023-04-19 16:03:30 26.0 Zero Values3 2023-04-20 05:16:30 2023-04-20 05:26:30 10.0 Zero Values4 2023-04-21 14:37:30 2023-04-21 14:49:30 12.0 Zero Values5 2023-04-22 23:07:30 2023-04-22 23:25:30 18.0 Zero Values6 2023-04-24 20:07:30 2023-04-24 20:26:30 19.0 Zero Values7 2023-04-25 11:07:30 2023-04-25 11:18:30 11.0 Zero Values...in the dataset, there are more instances which are not getting captured here.imageif you see only red block instances getting captured but after the first red block, in the next window the power is 0 continuously for greater than 10 minute