Skip to content

predict ¤

Prediction module, called after model training.

custom_predict ¤

custom_predict(X: Series, model: Pipeline, args: Namespace, y_true: Optional[ndarray] = None) -> tuple[ndarray, Namespace]

If the model has predict_proba attribute, predict the probability of each label occurring. Furthermore, if the true labels are given, use them to tune the threshold for each class using train.tune_threshold. Otherwise, if the model has no predict_proba attribute, predict the label directly (0 or 1) using 0.5 threshold.

Parameters:

  • X (Series) –

    Preprocessed posts.

  • model (Pipeline) –

    End-to-end pipeline including vectorizer and model.

  • args (Namespace) –

    Arguments containing booleans for preprocessing the posts and hyperparameters for the modeling pipeline. Can also contain the best threshold tuned for each class.

  • y_true (Optional[ndarray], default: None ) –

    Ground truth (correct) target values. Defaults to None.

Returns:

  • y_pred ( ndarray ) –

    Estimated targets as returned by the model.

  • args ( Namespace ) –

    Arguments, either is the same as input arguments or additionally also contains the best threshold tuned for each class.

Source code in tagolym/predict.py 
 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
def custom_predict(X: Series, model: Pipeline, args: Namespace, y_true: Optional[ndarray] = None) -> tuple[ndarray, Namespace]:
    """If the model has `predict_proba` attribute, predict the probability of 
    each label occurring. Furthermore, if the true labels are given, use them 
    to tune the threshold for each class using [train.tune_threshold][]. 
    Otherwise, if the model has no `predict_proba` attribute, predict the 
    label directly (0 or 1) using 0.5 threshold.

    Args:
        X (Series): Preprocessed posts.
        model (Pipeline): End-to-end pipeline including vectorizer and model.
        args (Namespace): Arguments containing booleans for preprocessing the 
            posts and hyperparameters for the modeling pipeline. Can also 
            contain the best threshold tuned for each class.
        y_true (Optional[ndarray], optional): Ground truth (correct) target 
            values. Defaults to None.

    Returns:
        y_pred: Estimated targets as returned by the model.
        args: Arguments, either is the same as input arguments or additionally 
            also contains the best threshold tuned for each class.
    """
    # prioritize predict_proba over predict
    if hasattr(model, "predict_proba"):
        y_score = model.predict_proba(X)
        y_score = np.array(y_score)[:, :, 1].T

        # tune threshold if label is given
        if y_true is not None:
            args.threshold = train.tune_threshold(y_true, y_score)

        y_pred = y_score > args.threshold
    else:
        y_pred = model.predict(X)
    return y_pred, args

predict ¤

predict(texts: list[str], artifacts: dict[str, Any]) -> list[dict]

Load arguments, label binarizer, and the trained model. Then, preprocess given posts and predict their labels using custom_predict. The label binarizer is used to transform the prediction matrix back into readable labels.

Parameters:

  • texts (list[str]) –

    User input list of posts.

  • artifacts (dict[str, Any]) –

    Arguments, label binarizer, and the trained model.

Returns:

  • list[dict] –

    List of key-value pairs of post and its label prediction.

Source code in tagolym/predict.py 
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
def predict(texts: list[str], artifacts: dict[str, Any]) -> list[dict]:
    """Load arguments, label binarizer, and the trained model. Then, 
    preprocess given posts and predict their labels using 
    [custom_predict][predict.custom_predict]. The label binarizer is used to 
    transform the prediction matrix back into readable labels.

    Args:
        texts (list[str]): User input list of posts.
        artifacts (dict[str, Any]): Arguments, label binarizer, and the 
            trained model.

    Returns:
        List of key-value pairs of post and its label prediction.
    """
    # load artifacts
    args = artifacts["args"]
    mlb = artifacts["label_encoder"]
    model = artifacts["model"]

    # predict
    x = pd.Series([data.preprocess_post(txt, args.nocommand, args.stem) for txt in texts])
    y, args = custom_predict(x, model, args)
    tags = mlb.inverse_transform(y)
    predictions = [
        {
            "input_text": texts[i],
            "predicted_tags": tags[i],
        }
        for i in range(len(texts))
    ]
    return predictions