minor: Support pandas 3.0 and numpy 2.4 (#2715)

Author: tcuongd

README.md

@@ -129,6 +129,44 @@ Using `cmdstanpy` with Windows requires a Unix-compatible C compiler such as min
 
 See [Release Notes](https://github.com/facebook/prophet/releases).
 
+### Version 1.3.0 (2026.01.27)
+
+#### Python
+
+- Support pandas>=3.0 and numpy>=2.4.
+
+### Version 1.2.2 (2026.01.25)
+
+#### Python
+
+- Version constraints on pandas (`<3`) and numpy (`<2.4`).
+
+#### R
+- Update build requirements to C++17 to Comply with CRAN Policy.
+- Add .tar.gz upload for R package to CI.
+- Re-generated holidays.csv for R package.
+
+### Version 1.2.1 (2025.10.22)
+
+#### Python
+
+- Also copy makefile to fake cmdstan.
+
+### Version 1.2.0 (2025.05.30)
+
+#### Python
+
+- Use latest CmdStan.
+- Add null check to CmdStanPyBackend cleanup() function.
+
+### Version 1.1.7 (2025.05.30)
+
+#### Python
+
+- Enable creation of custom performance metrics.
+- chore: address pandas futurewarning from "M" being deprecated.
+- cleanup() for cross_validate.
+
 ### Version 1.1.6 (2024.09.29)
 
 #### Python

python/prophet/__version__.py

@@ -1 +1 @@
-__version__ = "1.2.2"
+__version__ = "1.3.0"

python/prophet/diagnostics.py

@@ -10,10 +10,13 @@
 from tqdm.auto import tqdm
 from copy import deepcopy
 import concurrent.futures
+import multiprocessing
+import sys
 
 import numpy as np
 import pandas as pd
 
+
 logger = logging.getLogger('prophet')
 
 
@@ -107,7 +110,7 @@ def map(self, func, *iterables):
                         for args in zip(*iterables)
                      ]
                      return results
-                     
+
     disable_tqdm: if True it disables the progress bar that would otherwise show up when parallel=None
     extra_output_columns: A String or List of Strings e.g. 'trend' or ['trend'].
          Additional columns to 'yhat' and 'ds' to be returned in output.
@@ -116,27 +119,27 @@ def map(self, func, *iterables):
     -------
     A pd.DataFrame with the forecast, actual value and cutoff.
     """
-    
+
     if model.history is None:
         raise Exception('Model has not been fit. Fitting the model provides contextual parameters for cross validation.')
-    
+
     df = model.history.copy().reset_index(drop=True)
     horizon = pd.Timedelta(horizon)
     predict_columns = ['ds', 'yhat']
-        
+
     if model.uncertainty_samples:
         predict_columns.extend(['yhat_lower', 'yhat_upper'])
 
     if extra_output_columns is not None:
         if isinstance(extra_output_columns, str):
             extra_output_columns = [extra_output_columns]
         predict_columns.extend([c for c in extra_output_columns if c not in predict_columns])
-        
+
     # Identify the largest seasonality period
     period_max = 0.
     for s in model.seasonalities.values():
         period_max = max(period_max, s['period'])
-    seasonality_dt = pd.Timedelta(str(period_max) + ' days')    
+    seasonality_dt = pd.Timedelta(str(period_max) + ' days')
 
     if cutoffs is None:
         # Set period
@@ -152,15 +155,15 @@ def map(self, func, *iterables):
         cutoffs = generate_cutoffs(df, horizon, initial, period)
     else:
         # add validation of the cutoff to make sure that the min cutoff is strictly greater than the min date in the history
-        if min(cutoffs) <= df['ds'].min(): 
+        if min(cutoffs) <= df['ds'].min():
             raise ValueError("Minimum cutoff value is not strictly greater than min date in history")
         # max value of cutoffs is <= (end date minus horizon)
-        end_date_minus_horizon = df['ds'].max() - horizon 
-        if max(cutoffs) > end_date_minus_horizon: 
+        end_date_minus_horizon = df['ds'].max() - horizon
+        if max(cutoffs) > end_date_minus_horizon:
             raise ValueError("Maximum cutoff value is greater than end date minus horizon, no value for cross-validation remaining")
         initial = cutoffs[0] - df['ds'].min()
-        
-    # Check if the initial window 
+
+    # Check if the initial window
     # (that is, the amount of time between the start of the history and the first cutoff)
     # is less than the maximum seasonality period
     if initial < seasonality_dt:
@@ -175,7 +178,11 @@ def map(self, func, *iterables):
         if parallel == "threads":
             pool = concurrent.futures.ThreadPoolExecutor()
         elif parallel == "processes":
-            pool = concurrent.futures.ProcessPoolExecutor()
+            if sys.platform.startswith("win") or sys.platform == "darwin":
+                ctx = multiprocessing.get_context("spawn")
+            else:
+                ctx = multiprocessing.get_context("forkserver")
+            pool = concurrent.futures.ProcessPoolExecutor(mp_context=ctx)
         elif parallel == "dask":
             try:
                 from dask.distributed import get_client
@@ -204,7 +211,7 @@ def map(self, func, *iterables):
 
     else:
         predicts = [
-            single_cutoff_forecast(df, model, cutoff, horizon, predict_columns) 
+            single_cutoff_forecast(df, model, cutoff, horizon, predict_columns)
             for cutoff in (tqdm(cutoffs) if not disable_tqdm else cutoffs)
         ]
 
@@ -334,7 +341,7 @@ def register_performance_metric(func):
     df: Cross-validation results dataframe.
     w: Aggregation window size.
 
-    Registered metric should return following 
+    Registered metric should return following
     -------
     Dataframe with columns horizon and metric.
     """
@@ -382,7 +389,7 @@ def performance_metrics(df, metrics=None, rolling_window=0.1, monthly=False):
         use ['mse', 'rmse', 'mae', 'mape', 'mdape', 'smape', 'coverage'].
     rolling_window: Proportion of data to use in each rolling window for
         computing the metrics. Should be in [0, 1] to average.
-    monthly: monthly=True will compute horizons as numbers of calendar months 
+    monthly: monthly=True will compute horizons as numbers of calendar months
         from the cutoff date, starting from 0 for the cutoff month.
 
     Returns
@@ -477,7 +484,7 @@ def rolling_mean_by_h(x, h, w, name):
     res_x = res_x[(trailing_i + 1):]
 
     return pd.DataFrame({'horizon': res_h, name: res_x})
-    
+
 
 
 def rolling_median_by_h(x, h, w, name):

python/prophet/forecaster.py

@@ -27,7 +27,7 @@
 
 class Prophet(object):
     stan_backend: IStanBackend
-    
+
     """Prophet forecaster.
 
     Parameters
@@ -476,8 +476,8 @@ def fourier_series(
         if not (series_order >= 1):
             raise ValueError("series_order must be >= 1")
 
-        # convert to days since epoch
-        t = dates.to_numpy(dtype=np.int64) // NANOSECONDS_TO_SECONDS / (3600 * 24.)
+        epoch = pd.Timestamp("1970-01-01", tz=dates.dt.tz)
+        t = (dates - epoch).dt.total_seconds() / (24 * 60 * 60)
 
         x_T = t * np.pi * 2
         fourier_components = np.empty((dates.shape[0], 2 * series_order))
@@ -936,7 +936,7 @@ def add_group_component(self, components, name, group):
         group_cols = new_comp['col'].unique()
         if len(group_cols) > 0:
             new_comp = pd.DataFrame({'col': group_cols, 'component': name})
-            components = pd.concat([components, new_comp])
+            components = pd.concat([components, new_comp], ignore_index=True)
         return components
 
     def parse_seasonality_args(self, name, arg, auto_disable, default_order):
@@ -1332,16 +1332,19 @@ def piecewise_logistic(t, cap, deltas, k, m, changepoint_ts):
         Vector y(t).
         """
         # Compute offset changes
-        k_cum = np.concatenate((np.atleast_1d(k), np.cumsum(deltas) + k))
+        # Ensure k and m are scalars for numpy 2.x compatibility
+        k_scalar = np.asarray(k).item() if np.asarray(k).size == 1 else k
+        m_scalar = np.asarray(m).item() if np.asarray(m).size == 1 else m
+        k_cum = np.concatenate((np.atleast_1d(k_scalar), np.cumsum(deltas) + k_scalar))
         gammas = np.zeros(len(changepoint_ts))
         for i, t_s in enumerate(changepoint_ts):
             gammas[i] = (
-                    (t_s - m - np.sum(gammas))
+                    (t_s - m_scalar - np.sum(gammas))
                     * (1 - k_cum[i] / k_cum[i + 1])  # noqa W503
             )
         # Get cumulative rate and offset at each t
-        k_t = k * np.ones_like(t)
-        m_t = m * np.ones_like(t)
+        k_t = k_scalar * np.ones_like(t)
+        m_t = m_scalar * np.ones_like(t)
         for s, t_s in enumerate(changepoint_ts):
             indx = t >= t_s
             k_t[indx] += deltas[s]

python/prophet/serialize.py

@@ -169,7 +169,8 @@ def model_from_dict(model_dict):
                 s = s.dt.tz_localize(None)
             setattr(model, attribute, s)
     for attribute in PD_TIMESTAMP:
-        setattr(model, attribute, pd.Timestamp.utcfromtimestamp(model_dict[attribute]).tz_localize(None))
+        pd_ts = pd.Timestamp.fromtimestamp(model_dict[attribute], tz="UTC").tz_localize(None)
+        setattr(model, attribute, pd_ts)
     for attribute in PD_TIMEDELTA:
         setattr(model, attribute, pd.Timedelta(seconds=model_dict[attribute]))
     for attribute in PD_DATAFRAME:

python/prophet/tests/test_prophet.py

@@ -305,7 +305,7 @@ def test_flat_growth(self, backend, scaling):
         x = np.linspace(0, 2 * np.pi, 8 * 7)
         history = pd.DataFrame(
             {
-                "ds": pd.date_range(start="2020-01-01", periods=8 * 7, freq="d"),
+                "ds": pd.date_range(start="2020-01-01", periods=8 * 7, freq="D"),
                 "y": 30 + np.sin(x * 8.0),
             }
         )

python/prophet/tests/test_serialize.py

@@ -40,7 +40,10 @@ def test_simple_serialize(self, daily_univariate_ts, backend):
             elif k in PD_SERIES and v is not None:
                 assert v.equals(m2.__dict__[k])
             elif k in PD_DATAFRAME and v is not None:
-                pd.testing.assert_frame_equal(v, m2.__dict__[k], check_index_type=False)
+                # check_dtype=False since .fit() and .predict() will cooerce to the correct types
+                pd.testing.assert_frame_equal(
+                    v, m2.__dict__[k], check_index_type=False, check_dtype=False
+                )
             elif k == "changepoints_t":
                 assert np.array_equal(v, m.__dict__[k])
             else:
@@ -111,7 +114,10 @@ def test_full_serialize(self, daily_univariate_ts, backend):
             elif k in PD_SERIES and v is not None:
                 assert v.equals(m2.__dict__[k])
             elif k in PD_DATAFRAME and v is not None:
-                pd.testing.assert_frame_equal(v, m2.__dict__[k], check_index_type=False)
+                # check_dtype=False since .fit() and .predict() will cooerce to the correct types
+                pd.testing.assert_frame_equal(
+                    v, m2.__dict__[k], check_index_type=False, check_dtype=False
+                )
             elif k == "changepoints_t":
                 assert np.array_equal(v, m.__dict__[k])
             else:

python/pyproject.toml

@@ -14,9 +14,9 @@ readme = "README.md"
 requires-python = ">=3.7"
 dependencies = [
   "cmdstanpy>=1.0.4",
-  "numpy>=1.15.4,<2.4.0",
+  "numpy>=1.15.4",
   "matplotlib>=2.0.0",
-  "pandas>=1.0.4,<3",
+  "pandas>=1.0.4",
   "holidays>=0.25,<1",
   "tqdm>=4.36.1",
   "importlib_resources",
@@ -37,6 +37,7 @@ classifiers = [
   "Programming Language :: Python :: 3.9",
   "Programming Language :: Python :: 3.10",
   "Programming Language :: Python :: 3.11",
+  "Programming Language :: Python :: 3.12",
 ]
 
 [project.optional-dependencies]