add one in remaining place and remove ols check

Nixtla · Dec 19, 2023 · fdfb4b7 · fdfb4b7
1 parent 229ba2d
commit fdfb4b7
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 22 deletions.
diff --git a/nbs/src/arima.ipynb b/nbs/src/arima.ipynb
@@ -1217,12 +1217,9 @@
     "        if seasonal['period'] > 1 and seasonal['order'][1] > 0:\n",
     "            dx = diff(dx, seasonal['period'], seasonal['order'][1])\n",
     "            dxreg = diff(dxreg, seasonal['period'], seasonal['order'][1])\n",
-    "        if len(dx) > dxreg.shape[1]:\n",
-    "            model = sm.OLS(dx, dxreg)\n",
-    "            result = model.fit()\n",
-    "            fit = {'coefs': result.params, 'stderrs': result.bse}\n",
-    "        else:\n",
-    "            raise RuntimeError\n",
+    "        model = sm.OLS(dx, dxreg)\n",
+    "        result = model.fit()\n",
+    "        fit = {'coefs': result.params, 'stderrs': result.bse}\n",
     "        isna = np.isnan(x) | np.isnan(xreg).any(1)\n",
     "        n_used = (~isna).sum() - len(Delta)\n",
     "        init0 = np.append(init0, fit['coefs'])\n",
@@ -1755,7 +1752,7 @@
     "    missing = np.isnan(x)\n",
     "    missing_idxs = np.where(~missing)[0]\n",
     "    firstnonmiss = missing_idxs.min()\n",
-    "    lastnonmiss = missing_idxs.max()\n",
+    "    lastnonmiss = missing_idxs.max() + 1\n",
     "    n = np.sum(~missing[firstnonmiss:lastnonmiss])\n",
     "    m = seasonal['period']\n",
     "    seas_order = seasonal['order']\n",
@@ -3862,6 +3859,19 @@
     "#| hide\n",
     "AutoARIMA().fit(np.array([1]*36)).predict(20, level=80)"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "75f835a8-3459-4668-955c-2f1ed69aa7d4",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#| hide\n",
+    "number = 4.0\n",
+    "preds = AutoARIMA().fit(np.array([number])).predict(3)\n",
+    "np.testing.assert_array_equal(preds['mean'], number)"
+   ]
   }
  ],
  "metadata": {

diff --git a/statsforecast/arima.py b/statsforecast/arima.py
@@ -849,12 +849,9 @@ def maInvert(ma):
         if seasonal["period"] > 1 and seasonal["order"][1] > 0:
             dx = diff(dx, seasonal["period"], seasonal["order"][1])
             dxreg = diff(dxreg, seasonal["period"], seasonal["order"][1])
-        if len(dx) > dxreg.shape[1]:
-            model = sm.OLS(dx, dxreg)
-            result = model.fit()
-            fit = {"coefs": result.params, "stderrs": result.bse}
-        else:
-            raise RuntimeError
+        model = sm.OLS(dx, dxreg)
+        result = model.fit()
+        fit = {"coefs": result.params, "stderrs": result.bse}
         isna = np.isnan(x) | np.isnan(xreg).any(1)
         n_used = (~isna).sum() - len(Delta)
         init0 = np.append(init0, fit["coefs"])
@@ -1210,7 +1207,7 @@ def myarima(
     missing = np.isnan(x)
     missing_idxs = np.where(~missing)[0]
     firstnonmiss = missing_idxs.min()
-    lastnonmiss = missing_idxs.max()
+    lastnonmiss = missing_idxs.max() + 1
     n = np.sum(~missing[firstnonmiss:lastnonmiss])
     m = seasonal["period"]
     seas_order = seasonal["order"]

diff --git a/statsforecast/models.py b/statsforecast/models.py
@@ -4706,21 +4706,21 @@ def __init__(
     ):
         """TSB model.
 
-        Teunter-Syntetos-Babai: A modification of Croston's method that replaces the inter-demand 
+        Teunter-Syntetos-Babai: A modification of Croston's method that replaces the inter-demand
         intervals with the demand probability $d_t$, which is defined as follows.
 
         $$
         d_t = \\begin{cases}
-            1  & \\text{if demand occurs at time t} \\\ 
+            1  & \\text{if demand occurs at time t} \\\
             0  & \\text{otherwise.}
         \\end{cases}
         $$
 
-        Hence, the forecast is given by 
+        Hence, the forecast is given by
 
         $$\hat{y}_t= \hat{d}_t\hat{z_t}$$
 
-        Both $d_t$ and $z_t$ are forecasted using SES. The smooting paramaters of each may differ, 
+        Both $d_t$ and $z_t$ are forecasted using SES. The smooting paramaters of each may differ,
         like in the optimized Croston's method.
 
         References
@@ -4730,11 +4730,11 @@ def __init__(
         Parameters
         ----------
         alpha_d : float
-            Smoothing parameter for demand. 
+            Smoothing parameter for demand.
         alpha_p : float
-            Smoothing parameter for probability. 
-        alias : str  
-            Custom name of the model. 
+            Smoothing parameter for probability.
+        alias : str
+            Custom name of the model.
         prediction_intervals : Optional[ConformalIntervals]
             Information to compute conformal prediction intervals.
             By default, the model will compute the native prediction