Added a predict function to get predictions

piecewise_regression/main.py

@@ -867,14 +867,19 @@ def plot_data(self, **kwargs):
         """
         plt.scatter(self.xx, self.yy, **kwargs)
 
-    def plot_fit(self, **kwargs):
+    def predict(self, xx_predict):
         """
-        Plot the fitted model as a series of straight lines.
-        Passes any kwargs to the matplotlib plot function, e.g. color="red".
+        Predict y values from x values given the fitted model.
+        Returns predictions as a list of numbers. 
 
+        :param xx: Data series in x-axis.
+        :type xx: list of floats
         """
+
+        xx_predict = validate_list_of_numbers(var=xx_predict, var_name="x values to predict", min_length=1)
+
         if not self.best_muggeo:
-            print("Algorithm didn't converge. No fit to plot.")
+            print("Algorithm didn't converge. No model to use for prediction.")
         else:
             # Get the final results from the fitted model variables
             # Params are in terms of [intercept, alpha, betas, gammas]
@@ -886,15 +891,26 @@ def plot_fit(self, **kwargs):
             alpha_hat = final_params[1]
             beta_hats = final_params[2:2 + len(breakpoints)]
 
-            xx_plot = np.linspace(min(self.xx), max(self.xx), 100)
-
-            # Build the fit plot segment by segment. Betas are defined as
+            # Predict y values segment by segment. Betas are defined as
             # difference in gradient from previous section
-            yy_plot = intercept_hat + alpha_hat * xx_plot
+            yy_predict = intercept_hat + alpha_hat * xx_predict
             for bp_count in range(len(breakpoints)):
-                yy_plot += beta_hats[bp_count] * \
-                    np.maximum(xx_plot - breakpoints[bp_count], 0)
+                yy_predict += beta_hats[bp_count] * \
+                    np.maximum(xx_predict - breakpoints[bp_count], 0)
+            return yy_predict
+
+    def plot_fit(self, **kwargs):
+        """
+        Plot the fitted model as a series of straight lines.
+        Passes any kwargs to the matplotlib plot function, e.g. color="red".
 
+        """
+        if not self.best_muggeo:
+            print("Algorithm didn't converge. No fit to plot.")
+        else:
+            # Plot model
+            xx_plot = np.linspace(min(self.xx), max(self.xx), 100)
+            yy_plot = self.predict(xx_plot)
             plt.plot(xx_plot, yy_plot, **kwargs)
 
     def plot_breakpoints(self, **kwargs):