| """ |
| Weight predictor classes for food delivery platforms. |
| """ |
| import numpy as np |
|
|
|
|
| class WeightPredictor: |
| """Per-item-type weight predictor using TF-IDF + Ridge regression.""" |
| def __init__(self, tfidf, model): |
| self.tfidf = tfidf |
| self.model = model |
|
|
| def predict(self, texts): |
| """Predict weights for a list of texts.""" |
| X = self.tfidf.transform(texts) |
| return np.expm1(self.model.predict(X)) |
|
|
| def predict_single(self, text): |
| """Predict weight for a single text.""" |
| return self.predict([text])[0] |
|
|
|
|
| class UnifiedWeightPredictor: |
| """Unified predictor that routes to per-type models.""" |
| def __init__(self, predictors, default_type="grocery"): |
| self.predictors = predictors |
| self.default_type = default_type |
|
|
| def predict(self, text, item_type=None): |
| """ |
| Predict weight from text description. |
| item_type should be one of: menu_item, grocery, non_food |
| """ |
| if item_type is None: |
| if text.startswith("[MENU_ITEM]"): |
| item_type = "menu_item" |
| elif text.startswith("[GROCERY]"): |
| item_type = "grocery" |
| elif text.startswith("[NON_FOOD]"): |
| item_type = "non_food" |
| else: |
| item_type = self.default_type |
| predictor = self.predictors.get(item_type, self.predictors.get(self.default_type)) |
| return predictor.predict([text])[0] |
|
|
