PythonFiles/anomaly_detect.py at master · JEub/PythonFiles · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
import pandas as pd

def deviations_from_mean(data, values, group_by=None, window=5, sigma=1.96):
    """
    """
    # verifying data is correct instance
    assert isinstance(data,pd.DataFrame)

    ## checking that data entry is valid
    if group_by not in data.columns.tolist() and group_by is not None:
        raise KeyError('{} is not in the columns of the DataFrame object.'.format(group_by))
    if values not in data.columns.tolist():
        raise KeyError('{} is not in the columns of the DataFrame object.'.format(values))

    # Initializing return data
    df = pd.DataFrame()

    if group_by is not None:
        for name in data[group_by].unique().tolist():
            temp = data.loc[data[group_by]==name,:].copy()

            temp['rolling_mean'] = temp[values].rolling(window, min_periods=1).mean()
            temp['resid'] = temp[values] - temp['rolling_mean']

            ## computing the rolling standard deviation on the residuals
            temp['std'] = temp['resid'].rolling(window, min_periods=1).std()
            temp['std'].fillna(temp['std'].mean(),inplace=True)

            # defining lower and upper limits that would denote an abnormality
            temp['lower'] = temp['rolling_mean']-(temp['std']*sigma)
            temp['upper'] = temp['rolling_mean']+(temp['std']*sigma)

            temp['anomaly'] = 0
            temp.loc[(temp[values]<temp['lower'])|(temp[values]>temp['upper']),'anomaly'] = 1

            del temp['resid'], temp['lower'], temp['upper'], temp['rolling_mean'], temp['std']
            df = pd.concat([df,temp])
            del temp

    else:
        temp = data.copy()
        temp['rolling_mean'] = temp[values].rolling(window, min_periods=1).mean()
        temp['resid'] = temp[values] - temp['rolling_mean']

        ## computing the rolling standard deviation on the residuals
        temp['std'] = temp['resid'].rolling(window, min_periods=1).std()
        temp['std'].fillna(temp['std'].mean(),inplace=True)

        # defining lower and upper limits that would denote an abnormality
        temp['lower'] = temp['rolling_mean']-(temp['std']*sigma)
        temp['upper'] = temp['rolling_mean']+(temp['std']*sigma)

        temp['anomaly'] = 0
        temp.loc[(temp[values]<temp['lower'])|(temp[values]>temp['upper']),'anomaly'] = 1

        del temp['resid'], temp['lower'], temp['upper'], temp['rolling_mean'], temp['std']
        df = pd.concat([df,temp])
        del temp

    return df