make predict for single observation available in package

Author: watsang

Makefile

@@ -42,7 +42,7 @@ test-package:
 	coverage run -m pytest ./tests/
 	coverage report -m --include=./tests/*
 
-test: generate-dataset train prediction test-package ## run extensive tests
+test: generate-dataset train prediction clean test-package ## run extensive tests
 
 help: ## show help on available commands
 	@grep -E '^[a-zA-Z_-]+:.*?## .*$$' $(MAKEFILE_LIST) | sort | awk 'BEGIN {FS = ":.*?## "}; {printf "\033[36m%-30s\033[0m %s\n", $$1, $$2}'

ml_skeleton_py/model/__init__.py

@@ -2,9 +2,6 @@
 
 from .train import train
 
-from .predict import predict_from_file
+from .predict import predict_from_file, predict
 
-__all__ = [
-    "train",
-    "predict_from_file",
-]
+__all__ = ["train", "predict", "predict_from_file"]

ml_skeleton_py/model/predict.py

@@ -28,7 +28,7 @@ def load_model(model_name: str) -> BaseEstimator:
     Uses lru for caching.
 
     Parameters:
-        model_name (str): model name e.g. "lr.p"
+        model_name (str): model name (including extension) e.g. "lr.p"
 
     Returns:
         model (Pipeline or BaseEstimator): a model that can make predictions
@@ -38,14 +38,16 @@ def load_model(model_name: str) -> BaseEstimator:
     return model
 
 
-def predict(model_name: str, observation: np.array) -> float:
+def predict(observation: np.array, model_name: str = "lr.p") -> float:
     """
     Predict one single observation.
 
     Parameters:
-        model_name (str): the name of the model file that you want to use
+        observation (np.array): the input observation
 
-        observation (np.array): the variables of the input observation
+        model_name (str): the name of the model file that you want to load
+                          (including extension)
+                          default value = "lr.p"
 
     Return:
         prediction (float): the prediction
@@ -82,7 +84,7 @@ def predict_from_file(model: str, input_df: str, output_df: str) -> np.array:
     logger.info(f"running predictions for input: {input_data}")
 
     # make predictions
-    preds = [predict(model, [x]) for x in np.array(input_data)]
+    preds = [predict([x], model) for x in np.array(input_data)]
     # transform single axis array to a column
     preds = np.array(preds).reshape(-1, 1)
 

tests/test_predict.py

@@ -47,7 +47,7 @@ def test_predict_1() -> None:
         -1.25077894,
         -0.30741284,
     ]
-    prediction = predict(model_name, [observation])
+    prediction = predict([observation], model_name)
     assert type(float(prediction)) == float
 
 
@@ -88,7 +88,7 @@ def test_predict_2() -> None:
         -0.67713242,
         -0.20959966,
     ]
-    prediction = predict(model_name, [observation])
+    prediction = predict([observation], model_name)
     assert type(float(prediction)) == float
 
 
@@ -129,7 +129,7 @@ def test_predict_3() -> None:
         -1.08784478,
         3.03081115,
     ]
-    prediction = predict(model_name, [observation])
+    prediction = predict([observation], model_name)
     assert type(float(prediction)) == float