Add local tests

data/predict.py

@@ -37,7 +37,7 @@ def append_results(results):
 
 def main():
     predictor = initialize_predictor()
-    test_data = pd.read_csv(Constants.TEST_DATA_PATH)
+    test_data = pd.read_csv(Constants.TEST_DATA_PATH, header=None)
 
     for _, row in test_data.iterrows():
         result = predict_batch(predictor, [row])

evaluator.py

@@ -7,8 +7,7 @@ from sklearn.metrics import f1_score, log_loss
 
 class Constants:
     SHARED_DISK = os.getenv("AICROWD_SHARED_DIR", "test/shared")
-    PREDICTIONS_DIR = os.getenv("AICROWD_PREDICTIONS_DIR", "test/predictions")
-    GROUND_TRUTH_DIR = os.getenv("AICROWD_GROUND_TRUTH_DIR", "test")
+    GROUND_TRUTH_DIR = os.getenv("AICROWD_GROUND_TRUTH_DIR", "test/ground_truth")
 
     PREDICTIONS_FILE_PATH = os.path.join(SHARED_DISK, "predictions.csv")
     GROUND_TRUTH_PATH = os.path.join(GROUND_TRUTH_DIR, "test_ground_truth.csv")
@@ -27,7 +26,7 @@ class AIcrowdEvaluator:
 
         labels = [i for i in range(0, 10)]
         y_pred = np.zeros((len(predictions), len(labels)))
-        for index, val in enumerate(submission):
+        for index, val in enumerate(predictions):
             y_pred[index][val] = 1
  
 
@@ -39,3 +38,9 @@ class AIcrowdEvaluator:
             "score_secondary": log_loss_score,
         }
 
+
+if __name__ == "__main__":
+    evaluator = AIcrowdEvaluator()
+    score = evaluator.evaluate()
+    print(score)
+

test/ground_truth/test_ground_truth.csv

+2
+3
+6
+1
+2
+4
+6
+9
+0

test/shared/predictions.csv

+1
+2
+3
+4
+5
+6
+7
+8
+9