Remove pystan, leave backend abstraction for now (#2148)

.github/workflows/wheel.yml

@@ -69,7 +69,7 @@ jobs:
           CIBW_ENVIRONMENT: >
             STAN_BACKEND="${{ env.STAN_BACKEND }}"
             PIP_CACHE_DIR="${{ env.PIP_DEFAULT_CACHE }}"
-          CIBW_BUILD: cp36-* cp37-* cp38-*
+          CIBW_BUILD: cp36-* cp37-* cp38-* cp39-* cp310-*
           CIBW_ARCHS: native
           CIBW_BUILD_FRONTEND: build
           # CIBW_REPAIR_WHEEL_COMMAND: delvewheel repair -w {dest_dir} {wheel}
@@ -142,7 +142,7 @@ jobs:
             PIP_CACHE_DIR="/host/${{ env.PIP_DEFAULT_CACHE }}"
           CIBW_MANYLINUX_X86_64_IMAGE: manylinux2014
           CIBW_BEFORE_ALL_LINUX: chmod -R a+rwx /host/${{ env.PIP_DEFAULT_CACHE }}
-          CIBW_BUILD: cp36-* cp37-* cp38-*
+          CIBW_BUILD: cp36-* cp37-* cp38-* cp39-* cp310-*
           CIBW_SKIP: "*musllinux*"
           CIBW_ARCHS: native
           CIBW_BUILD_FRONTEND: build

README.md

@@ -78,13 +78,6 @@ $ CMDSTAN=/tmp/cmdstan-2.22.1 STAN_BACKEND=CMDSTANPY pip install prophet
 
 Note that the `CMDSTAN` variable is directly related to `cmdstanpy` module and can be omitted if your CmdStan binaries are in your `$PATH`.
 
-It is also possible to install Prophet with two backends:
-
-```bash
-# bash
-$ CMDSTAN=/tmp/cmdstan-2.22.1 STAN_BACKEND=PYSTAN,CMDSTANPY pip install prophet
-```
-
 After installation, you can [get started!](https://facebook.github.io/prophet/docs/quick_start.html#python-api)
 
 If you upgraded the version of PyStan installed on your system, you may need to reinstall prophet ([see here](https://github.com/facebook/prophet/issues/324)).

docs/_docs/uncertainty_intervals.md

@@ -73,12 +73,8 @@ prophet_plot_components(m, forecast)
 # Python
 fig = m.plot_components(forecast)
 ```
- 
-![png](/prophet/static/uncertainty_intervals_files/uncertainty_intervals_11_0.png) 
+
+![png](/prophet/static/uncertainty_intervals_files/uncertainty_intervals_11_0.png)
 
 
 You can access the raw posterior predictive samples in Python using the method `m.predictive_samples(future)`, or in R using the function `predictive_samples(m, future)`.
-
-
-There are upstream issues in PyStan for Windows which make MCMC sampling extremely slow. The best choice for MCMC sampling in Windows is to use R, or Python in a Linux VM.
-

python/prophet/models.py

@@ -227,70 +227,9 @@ class CmdStanPyBackend(IStanBackend):
         return output
 
 
-class PyStanBackend(IStanBackend):
-
-    @staticmethod
-    def get_type():
-        return StanBackendEnum.PYSTAN.name
-
-    def sampling(self, stan_init, stan_data, samples, **kwargs) -> dict:
-
-        args = dict(
-            data=stan_data,
-            init=lambda: stan_init,
-            iter=samples,
-        )
-        args.update(kwargs)
-        self.stan_fit = self.model.sampling(**args)
-        out = {}
-        for par in self.stan_fit.model_pars:
-            out[par] = self.stan_fit[par]
-            # Shape vector parameters
-            if par in ['delta', 'beta'] and len(out[par].shape) < 2:
-                out[par] = out[par].reshape((-1, 1))
-        return out
-
-    def fit(self, stan_init, stan_data, **kwargs) -> dict:
-
-        args = dict(
-            data=stan_data,
-            init=lambda: stan_init,
-            algorithm='Newton' if stan_data['T'] < 100 else 'LBFGS',
-            iter=1e4,
-        )
-        args.update(kwargs)
-        try:
-            self.stan_fit = self.model.optimizing(**args)
-        except RuntimeError as e:
-            # Fall back on Newton
-            if self.newton_fallback and args['algorithm'] != 'Newton':
-                logger.warning(
-                    'Optimization terminated abnormally. Falling back to Newton.'
-                )
-                args['algorithm'] = 'Newton'
-                self.stan_fit = self.model.optimizing(**args)
-            else:
-                raise e
-
-        params = {}
-
-        for par in self.stan_fit.keys():
-            params[par] = self.stan_fit[par].reshape((1, -1))
-
-        return params
-
-    def load_model(self):
-        """Load compiled Stan model"""
-        model_file = pkg_resources.resource_filename(
-            'prophet',
-            'stan_model/prophet_model.pkl',
-        )
-        with Path(model_file).open('rb') as f:
-            return pickle.load(f)
 
 
 class StanBackendEnum(Enum):
-    PYSTAN = PyStanBackend
     CMDSTANPY = CmdStanPyBackend
 
     @staticmethod

python/requirements.txt

@@ -1,6 +1,5 @@
 Cython>=0.22
-cmdstanpy>=1.0.0
-pystan~=2.19.1.1
+cmdstanpy>=1.0.1
 numpy>=1.15.4
 pandas>=1.0.4
 matplotlib>=2.0.0

python/setup.py

@@ -118,34 +118,17 @@ def build_cmdstan_model(target_dir):
         prune_cmdstan(cmdstan_dir)
 
 
-def build_pystan_model(target_dir):
-    """
-    Compile the stan model using pystan and pickle it. The pickle is copied to {target_dir}/prophet_model.pkl.
-    """
-    import pystan
-    import pickle
-
-    model_name = "prophet.stan"
-    target_name = "prophet_model.pkl"
-    with open(os.path.join(MODEL_DIR, model_name)) as f:
-        model_code = f.read()
-    sm = pystan.StanModel(model_code=model_code)
-    with open(os.path.join(target_dir, target_name), "wb") as f:
-        pickle.dump(sm, f, protocol=pickle.HIGHEST_PROTOCOL)
-
 
 def get_backends_from_env() -> List[str]:
     return os.environ.get("STAN_BACKEND", "CMDSTANPY").split(",")
 
 
 def build_models(target_dir):
-    for backend in get_backends_from_env():
-        print(f"Compiling {backend} model")
-        if backend == "CMDSTANPY":
-            build_cmdstan_model(target_dir)
-        elif backend == "PYSTAN" and PLATFORM != "win":
-            build_pystan_model(target_dir)
+    print(f"Compiling cmdstanpy model")
+    build_cmdstan_model(target_dir)
 
+    if 'PYSTAN' in get_backends_from_env():
+        raise ValueError("PyStan backend is not supported for Prophet >= 1.1")
 
 class BuildPyCommand(build_py):
     """Custom build command to pre-compile Stan models."""