- add notebook that runs the plotting code
- add three visualizations per city:
  + all addresses, colored by zip code
  + all restaurants, incl. the number of received orders
  + all restaurants on a grid with pixel side length of 1000m

... after upgrading:
- alembic
- matplotlib
- pandas
- rpy2
- sqlalchemy
- statsmodels
- dev dependencies
  + coverage     + factory-boy    + faker                 + nox
  + packaging    + pre-commit     + flake8-annotations    + pytest
  + pytest-cov   + sphinx
- research dependencies
  + numpy        + pyty
- transient dependencies
  + astpretty    + atomicwrites   + bleach                + chardet
  + colorlog     + darglint       + flake8-comprehensions + gitpython
  + identify     + ipykernel      + ipython               + jedi
  + jinja2       + jupyter-client + jupyter-core          + mako
  + nbformat     + nest-asyncio   + notebook              + parso
  + pluggy       + prompt-toolkit + ptyprocess            + pygments
  + pyyaml       + pyzmq          + requests              + smmap
  + terminado    + textfixtures   + snowballstemmer       + typed-ast
  + urllib3      + virtualenv

- fix SQL statements written in raw text

- use the `HorizontalSMAModel` for low demand
- use the `TrivialModel` for no demand

- we use that shorter name in `urban_meal_delivery.forecasts.*`
  and want to be consistent in the ORM layer as well

- the trivial model simply predicts `0` demand for all time steps

- the model applies a simple moving average on horizontal time series
- refactor `db.Forecast.from_dataframe()` to correctly convert
  `float('NaN')` values into `None`; otherwise, SQLAlchemy complains

- the method implements a heuristic from the first research paper
  that chooses the most promising forecasting `*Model` based on
  the average daily demand in a `Pixel` for a given `train_horizon`
- adjust the test scenario => `LONG_TRAIN_HORIZON` becomes `8`
  as that is part of the rule implemented in the heuristic

- when running tests marked with "r" we still must not run tests
  marked with "db" on the CI server

- the method calculates the number of daily `Order`s in a `Pixel`
  withing the `train_horizon` preceding the `predict_day`

- `*Model`s use the `methods.*.predict()` functions to predict demand
  given an order time series generated by `timify.OrderHistory`
- `models.base.ForecastingModelABC` unifies how all `*Model`s work
  and implements a caching strategy
- implement three `*Model`s for tactical forecasting, based on the
  hets, varima, and rtarima models described in the first research paper
- add overall documentation for `urban_meal_delivery.forecasts` package
- move the fixtures in `tests.forecasts.timify.conftest` to
  `tests.forecasts.conftest` and adjust the horizon of the test horizon
  from two to three weeks

- this alternative constructor takes the `pd.DataFrame`s from the
  `*Model.predict()` methods and converts them into ORM models

- the function implements a forecasting "method" similar to the
  seasonal naive method
  => instead of simply taking the last observation given a seasonal lag,
     it linearly extrapolates all observations of the same seasonal lag
     from the past into the future; conceptually, it is like the
     seasonal naive method with built-in smoothing
- the function is tested just like the `arima.predict()` and
  `ets.predict()` functions
  + rename the `tests.forecasts.methods.test_ts_methods` module
    into `tests.forecasts.methods.test_predictions`
- re-organize some constants in the `tests` package
- streamline some docstrings

- get the `datetime` of the first or last order within a pixel
- unify some fixtures in `tests.forecasts.timify.conftest`

- rename "total_orders" columns into "n_orders"
- rename `.make_*_time_series()` methods into `.make_*_ts()`

- the property loads all `Restaurant`s from the database that
  are within the `Pixel`

- move the module
- unify the corresponding tests in `tests.forecasts.methods` sub-package
- make all `predict()` and the `stl()` function round results
- streamline documentation

- this code is not unit-tested due to the complexity involving
  interactive `folium.Map`s => visual checks give high confidence

- the properties are needed for the drawing functionalitites

- ensure a `Restaurant` only has one unique `Order.pickup_address`
- rework `Grid.gridify()` so that only pickup addresses are assigned
  into `Pixel`s
- include database migrations to ensure the data adhere to these
  tighter constraints

- this is a minor sanity check

- add `.low80`, `.high80`, `.low95`, and `.high95` columns
- add check contraints for the confidence intervals
- rename the `.method` column into `.model` for consistency

- `stl()` wraps R's "stl" function in Python
- STL is a decomposition method for time series