# 1. Description

The model for determining the spot price of transportation from point A to point B (only for direct routes).

The model is based on geographical coordinates, prices are predicted in rubles. 

Prices are valid only for transportation across the territory of the Russian Federation, the affiliation of specific coordinates to the territory of the Russian Federation is poorly controlled (see the config/dataset.yaml file, *coordinates_thr* parameter), and remains on the conscience of the user of the model.

# 2. Installation

    pip install git+https://gitea.dot-dot.ru/dot-dot/transit_calculator.git

# 3. Usage

### Load the model

```python
from tcalc.model import TCalcPredictor
from tcalc.config import TCalcDatasetConfig


predictor = TCalcPredictor(dataset_config=TCalcDatasetConfig())

# predictor can load models from links
predictor.load_models("https://storage.yandexcloud.net/data-monsters/sber/actual_model.zip")

# or from local folders (models will be loaded as ensemble)
# predictor.load_models("default_model")

# and files (single model)
# predictor.load_models("default_model/default_model.onnx")

```


### Inference

```python
data = [
        {
            "created_at": "2024-01-01 12:03:03", # when the order is created, i.e. the moment of determining the price
            "pick_at": "2024-01-10 09:00:00", # when the cargo needs to be picked up
            # geographical coordinates of loading and unloading locations
            "from_lat": 55.751,
            "from_lon": 37.618,
            "to_lat": 52.139,
            "to_lon": 104.21,
            # load capacity and volume of the truck
            "weight": 20000,
            "volume": 82,

            # 2 - refrigerator, 1 - all other types, 1 by default and can be skipped
            "car_type_id": 1
        }
    ]

output = predictor(data)
print(output)
print(output.shape)

>> [125048.414]
>> (1,)
```