fix: refactor backend

.github/workflows/ci.yml

@@ -105,10 +105,10 @@ jobs:
         run: uv sync --dev
 
       - name: Ruff lint
-        run: uv run ruff check src/
+        run: uv run ruff check foresight/
 
       - name: Ruff format check
-        run: uv run ruff format --check src/
+        run: uv run ruff format --check foresight/
 
   test:
     name: Tests

README.md

@@ -75,19 +75,21 @@ Commit messages must follow [Conventional Commits](https://www.conventionalcommi
 ```
 foresight/
 ├── backend/
-│   ├── src/
-│   │   └── foresight/           # Installable package (see pyproject.toml)
-│   │       ├── main.py          # FastAPI app entrypoint (foresight.main:app)
-│   │       ├── config.py        # Settings (pydantic-settings)
-│   │       ├── routes/          # API route handlers
-│   │       ├── schemas/         # Pydantic request/response models
-│   │       ├── services/        # Business logic
-│   │       ├── repository/      # Data-access layer
-│   │       └── database/        # SQLAlchemy + Alembic
-│   │           ├── base.py      # Declarative base
-│   │           ├── session.py   # Async session factory
-│   │           ├── models/      # SQLAlchemy models
-│   │           └── migrations/  # Alembic migrations
+│   ├── foresight/               # Installable package (see pyproject.toml)
+│   │   ├── main.py              # FastAPI app entrypoint (foresight.main:app)
+│   │   ├── config.py            # Settings (pydantic-settings)
+│   │   ├── routes/              # API route handlers
+│   │   ├── schemas/             # Pydantic request/response models
+│   │   ├── services/            # Business logic
+│   │   │   ├── ml/              # ML models (regression, bayesian, ...)
+│   │   │   └── parsers/         # Client data ingestion
+│   │   ├── repository/          # Data-access layer
+│   │   └── database/            # SQLAlchemy + Alembic
+│   │       ├── base.py          # Declarative base
+│   │       ├── session.py       # Async session factory
+│   │       ├── models/          # SQLAlchemy models
+│   │       └── migrations/      # Alembic migrations
+│   ├── scripts/                 # Dev tooling (mock data, model comparison)
 │   ├── tests/                   # Pytest suite
 │   ├── alembic.ini
 │   ├── pyproject.toml

backend/Dockerfile

@@ -30,7 +30,7 @@ EXPOSE 8000
 
 # ---- dev -------------------------------------------------------------------
 # Adds dev dependencies (pytest, ruff, ...) and runs uvicorn with --reload.
-# docker-compose.yml bind-mounts ./backend/src over /app/src so edits are
+# docker-compose.yml bind-mounts ./backend/foresight over /app/foresight so edits are
 # picked up live. Selected via `build.target: dev` in docker-compose.yml.
 FROM base AS dev
 

backend/alembic.ini

@@ -1,5 +1,5 @@
 [alembic]
-script_location = src/foresight/database/migrations
+script_location = foresight/database/migrations
 # The `foresight` package is installed via `uv sync`, so no sys.path
 # manipulation is needed for `from foresight...` imports in env.py to resolve.
 

backend/src/services/ml/bayesian.py → backend/foresight/services/ml/bayesian.py

@@ -3,7 +3,7 @@
 from sklearn.pipeline import make_pipeline
 from sklearn.preprocessing import StandardScaler
 
-from services.ml.base import Model
+from foresight.services.ml.base import Model
 
 
 class BayesianModel(Model):

backend/src/services/ml/neural_net.py → backend/foresight/services/ml/neural_net.py

@@ -2,7 +2,7 @@
 from sklearn.pipeline import make_pipeline
 from sklearn.preprocessing import StandardScaler
 
-from services.ml.base import Model
+from foresight.services.ml.base import Model
 
 DEFAULTS = {
     "hidden_layer_sizes": (32, 16),

backend/src/services/ml/regression.py → backend/foresight/services/ml/regression.py

@@ -1,7 +1,7 @@
 import numpy as np
 from sklearn.ensemble import HistGradientBoostingRegressor
 
-from services.ml.base import Model
+from foresight.services.ml.base import Model
 
 
 class GBMModel(Model):

backend/src/services/ml/timeseries.py → backend/foresight/services/ml/timeseries.py

@@ -1,6 +1,6 @@
 from statsmodels.tsa.statespace.sarimax import SARIMAX
 
-from services.ml.base import Model
+from foresight.services.ml.base import Model
 
 
 class TimeSeriesModel(Model):

backend/pyproject.toml

@@ -42,12 +42,12 @@ ml = [
 ]
 
 [tool.hatch.build.targets.wheel]
-packages = ["src/foresight"]
+packages = ["foresight"]
 
 [tool.ruff]
 target-version = "py312"
-src = ["src"]
-exclude = ["src/foresight/database/migrations"]
+src = ["."]
+exclude = ["foresight/database/migrations"]
 line-length = 88
 
 [tool.ruff.lint]
@@ -69,4 +69,4 @@ known-first-party = ["foresight"]
 [tool.pytest.ini_options]
 asyncio_mode = "auto"
 testpaths = ["tests"]
-pythonpath = ["src", "scripts"]
+pythonpath = ["scripts"]

backend/scripts/compare_models.py

@@ -3,19 +3,14 @@
 Run from backend/: uv run python scripts/compare_models.py
 """
 
-import sys
-from pathlib import Path
-
 import numpy as np
 import pandas as pd
-
 from generate_mock_data import write_csv
 
-sys.path.insert(0, str(Path(__file__).resolve().parents[1] / "src"))
-from services.ml.bayesian import BayesianModel
-from services.ml.neural_net import NNModel
-from services.ml.regression import GBMModel
-from services.ml.timeseries import TimeSeriesModel
+from foresight.services.ml.bayesian import BayesianModel
+from foresight.services.ml.neural_net import NNModel
+from foresight.services.ml.regression import GBMModel
+from foresight.services.ml.timeseries import TimeSeriesModel
 
 # Known only after the night, so never used to predict price.
 POST_STAY_COLUMNS = ["rooms_sold", "occupancy", "room_revenue"]

backend/tests/test_ml_models.py

@@ -4,11 +4,11 @@
 import pandas as pd
 import pytest
 
-from services.ml.base import Model
-from services.ml.bayesian import BayesianModel
-from services.ml.neural_net import NNModel
-from services.ml.regression import GBMModel
-from services.ml.timeseries import TimeSeriesModel
+from foresight.services.ml.base import Model
+from foresight.services.ml.bayesian import BayesianModel
+from foresight.services.ml.neural_net import NNModel
+from foresight.services.ml.regression import GBMModel
+from foresight.services.ml.timeseries import TimeSeriesModel
 
 MODELS = [BayesianModel, NNModel, GBMModel, TimeSeriesModel]
 
@@ -65,6 +65,7 @@ def test_nn_has_no_interval(price_split):
         model.predict_interval(x_test)
     assert model.evaluate(x_test, y_test)["coverage"] is None
 
+
 class FixedModel(Model):
     """Returns preset predictions and a +/-5 band, recording the alpha asked for."""