![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F18178120%2F3255a70467843ed9f4e8d49e3bbb4975%2FScreenshot%202025-12-23%20220013.png?generation=1766502225920231&alt=media\")
My submission was error. It has a problem with format but i can't fix it:
\n
def create_submission(test_csv, predictions_map):\n \"\"\"\n Create submission file in correct format\n CRITICAL: Maintains exact order and count from test.csv\n \"\"\"\n print(\"=\"*60)\n print(\"CREATING SUBMISSION\")\n print(\"=\"*60)\n\n # Load test CSV\n test_df = pd.read_csv(test_csv)\n print(f\"Test samples: {len(test_df)}\")\n\n # Build submission row by row (preserves order)\n results = []\n missing_count = 0\n\n for _, row in test_df.iterrows():\n img_path = row['image_path']\n target_name = row['target_name']\n sample_id = row['sample_id']\n\n # Get prediction for this specific target\n if img_path in predictions_map:\n pred_vector = predictions_map[img_path] # [5 values]\n target_idx = config.TARGET_TO_IDX[target_name]\n value = max(0.0, float(pred_vector[target_idx]))\n else:\n value = 0.0\n missing_count += 1\n\n results.append({\n 'sample_id': sample_id,\n 'target': value\n })\n\n # Create DataFrame\n submission = pd.DataFrame(results)\n\n # Warnings\n if missing_count > 0:\n print(f\"⚠ WARNING: {missing_count} samples had no predictions (filled with 0.0)\")\n\n return submission\n\n\ndef validate_submission(submission, test_csv):\n \"\"\"Validate submission format\"\"\"\n print(\"\\n\" + \"=\"*60)\n print(\"VALIDATION\")\n print(\"=\"*60)\n\n test_df = pd.read_csv(test_csv)\n\n checks = []\n\n # Check 1: Row count\n row_match = len(submission) == len(test_df)\n checks.append((\"Row count\", row_match, f\"{len(submission)} == {len(test_df)}\"))\n\n # Check 2: Columns\n col_match = list(submission.columns) == ['sample_id', 'target']\n checks.append((\"Columns\", col_match, str(submission.columns.tolist())))\n\n # Check 3: No nulls\n no_nulls = submission.isnull().sum().sum() == 0\n checks.append((\"No nulls\", no_nulls, f\"{submission.isnull().sum().sum()} nulls\"))\n\n # Check 4: Non-negative values\n non_negative = (submission['target'] >= 0).all()\n checks.append((\"Non-negative\", non_negative, f\"Min: {submission['target'].min():.4f}\"))\n\n # Check 5: Sample ID order\n order_match = (submission['sample_id'].values == test_df['sample_id'].values).all()\n checks.append((\"Order preserved\", order_match, \"Sample IDs match order\"))\n\n # Print results\n all_passed = True\n for name, passed, detail in checks:\n status = \"✓\" if passed else \"✗\"\n print(f\"{status} {name}: {detail}\")\n if not passed:\n all_passed = False\n\n if not all_passed:\n raise ValueError(\"Submission validation failed!\")\n\n # Statistics\n print(\"\\n\" + \"-\"*60)\n print(\"STATISTICS\")\n print(\"-\"*60)\n print(f\"Shape: {submission.shape}\")\n print(f\"Min: {submission['target'].min():.6f}\")\n print(f\"Max: {submission['target'].max():.6f}\")\n print(f\"Mean: {submission['target'].mean():.6f}\")\n print(f\"Median: {submission['target'].median():.6f}\")\n print(f\"Std: {submission['target'].std():.6f}\")\n\n print(\"\\n✓ All validation checks passed!\\n\")\n\n\ndef save_submission(submission, output_path):\n \"\"\"Save submission with proper formatting\"\"\"\n submission.to_csv(\n output_path, \n index=False, \n encoding='utf-8',\n float_format='%.6f'\n )\n\n # Verify saved file\n saved = pd.read_csv(output_path)\n assert saved.shape == submission.shape, \"Saved file shape mismatch!\"\n assert (saved['sample_id'] == submission['sample_id']).all(), \"Sample IDs changed!\"\n\n print(f\"✓ Submission saved: {output_path}\")\n print(f\" Shape: {saved.shape}\")\n print(f\" First row: {saved.iloc[0].to_dict()}\")\n\n\n# ==================== MAIN ====================\ndef run_inference():\n \"\"\"Main inference pipeline\"\"\"\n try:\n print(\"\\n\" + \"=\"*60)\n print(\"BIOMASS PREDICTION INFERENCE\")\n print(\"=\"*60 + \"\\n\")\n\n # Step 1: Load model\n model = load_model()\n\n # Step 2: Prepare data\n dataset = TestDataset(config.TEST_CSV, config.ROOT_DIR)\n dataloader = DataLoader(\n dataset,\n batch_size=config.BATCH_SIZE,\n shuffle=False,\n num_workers=2,\n pin_memory=True\n )\n\n # Step 3: Run inference with TTA\n predictions_map = predict_with_tta(model, dataloader)\n\n # Step 4: Create submission\n submission = create_submission(config.TEST_CSV, predictions_map)\n\n # Step 5: Validate\n validate_submission(submission, config.TEST_CSV)\n\n # Step 6: Save\n save_submission(submission, config.SUBMISSION_FILE)\n\n print(\"\\n\" + \"=\"*60)\n print(\"✓ INFERENCE COMPLETED SUCCESSFULLY\")\n print(\"=\"*60 + \"\\n\")\n\n except Exception as e:\n print(f\"\\n❌ ERROR: {e}\")\n import traceback\n traceback.print_exc()\n\n # Emergency fallback submission\n print(\"\\nGenerating fallback submission (all zeros)...\")\n try:\n test_df = pd.read_csv(config.TEST_CSV)\n fallback = pd.DataFrame({\n 'sample_id': test_df['sample_id'],\n 'target': 0.0\n })\n fallback.to_csv(config.SUBMISSION_FILE, index=False)\n print(f\"✓ Fallback saved: {len(fallback)} rows\")\n except Exception as e2:\n print(f\"❌ Fallback failed: {e2}\")\n raise\n\n\nif __name__ == \"__main__\":\n run_inference()\nMy submission was error. It has a problem with format but i can't fix it:
\n
def create_submission(test_csv, predictions_map):\n \"\"\"\n Create submission file in correct format\n CRITICAL: Maintains exact order and count from test.csv\n \"\"\"\n print(\"=\"*60)\n print(\"CREATING SUBMISSION\")\n print(\"=\"*60)\n\n # Load test CSV\n test_df = pd.read_csv(test_csv)\n print(f\"Test samples: {len(test_df)}\")\n\n # Build submission row by row (preserves order)\n results = []\n missing_count = 0\n\n for _, row in test_df.iterrows():\n img_path = row['image_path']\n target_name = row['target_name']\n sample_id = row['sample_id']\n\n # Get prediction for this specific target\n if img_path in predictions_map:\n pred_vector = predictions_map[img_path] # [5 values]\n target_idx = config.TARGET_TO_IDX[target_name]\n value = max(0.0, float(pred_vector[target_idx]))\n else:\n value = 0.0\n missing_count += 1\n\n results.append({\n 'sample_id': sample_id,\n 'target': value\n })\n\n # Create DataFrame\n submission = pd.DataFrame(results)\n\n # Warnings\n if missing_count > 0:\n print(f\"⚠ WARNING: {missing_count} samples had no predictions (filled with 0.0)\")\n\n return submission\n\n\ndef validate_submission(submission, test_csv):\n \"\"\"Validate submission format\"\"\"\n print(\"\\n\" + \"=\"*60)\n print(\"VALIDATION\")\n print(\"=\"*60)\n\n test_df = pd.read_csv(test_csv)\n\n checks = []\n\n # Check 1: Row count\n row_match = len(submission) == len(test_df)\n checks.append((\"Row count\", row_match, f\"{len(submission)} == {len(test_df)}\"))\n\n # Check 2: Columns\n col_match = list(submission.columns) == ['sample_id', 'target']\n checks.append((\"Columns\", col_match, str(submission.columns.tolist())))\n\n # Check 3: No nulls\n no_nulls = submission.isnull().sum().sum() == 0\n checks.append((\"No nulls\", no_nulls, f\"{submission.isnull().sum().sum()} nulls\"))\n\n # Check 4: Non-negative values\n non_negative = (submission['target'] >= 0).all()\n checks.append((\"Non-negative\", non_negative, f\"Min: {submission['target'].min():.4f}\"))\n\n # Check 5: Sample ID order\n order_match = (submission['sample_id'].values == test_df['sample_id'].values).all()\n checks.append((\"Order preserved\", order_match, \"Sample IDs match order\"))\n\n # Print results\n all_passed = True\n for name, passed, detail in checks:\n status = \"✓\" if passed else \"✗\"\n print(f\"{status} {name}: {detail}\")\n if not passed:\n all_passed = False\n\n if not all_passed:\n raise ValueError(\"Submission validation failed!\")\n\n # Statistics\n print(\"\\n\" + \"-\"*60)\n print(\"STATISTICS\")\n print(\"-\"*60)\n print(f\"Shape: {submission.shape}\")\n print(f\"Min: {submission['target'].min():.6f}\")\n print(f\"Max: {submission['target'].max():.6f}\")\n print(f\"Mean: {submission['target'].mean():.6f}\")\n print(f\"Median: {submission['target'].median():.6f}\")\n print(f\"Std: {submission['target'].std():.6f}\")\n\n print(\"\\n✓ All validation checks passed!\\n\")\n\n\ndef save_submission(submission, output_path):\n \"\"\"Save submission with proper formatting\"\"\"\n submission.to_csv(\n output_path, \n index=False, \n encoding='utf-8',\n float_format='%.6f'\n )\n\n # Verify saved file\n saved = pd.read_csv(output_path)\n assert saved.shape == submission.shape, \"Saved file shape mismatch!\"\n assert (saved['sample_id'] == submission['sample_id']).all(), \"Sample IDs changed!\"\n\n print(f\"✓ Submission saved: {output_path}\")\n print(f\" Shape: {saved.shape}\")\n print(f\" First row: {saved.iloc[0].to_dict()}\")\n\n\n# ==================== MAIN ====================\ndef run_inference():\n \"\"\"Main inference pipeline\"\"\"\n try:\n print(\"\\n\" + \"=\"*60)\n print(\"BIOMASS PREDICTION INFERENCE\")\n print(\"=\"*60 + \"\\n\")\n\n # Step 1: Load model\n model = load_model()\n\n # Step 2: Prepare data\n dataset = TestDataset(config.TEST_CSV, config.ROOT_DIR)\n dataloader = DataLoader(\n dataset,\n batch_size=config.BATCH_SIZE,\n shuffle=False,\n num_workers=2,\n pin_memory=True\n )\n\n # Step 3: Run inference with TTA\n predictions_map = predict_with_tta(model, dataloader)\n\n # Step 4: Create submission\n submission = create_submission(config.TEST_CSV, predictions_map)\n\n # Step 5: Validate\n validate_submission(submission, config.TEST_CSV)\n\n # Step 6: Save\n save_submission(submission, config.SUBMISSION_FILE)\n\n print(\"\\n\" + \"=\"*60)\n print(\"✓ INFERENCE COMPLETED SUCCESSFULLY\")\n print(\"=\"*60 + \"\\n\")\n\n except Exception as e:\n print(f\"\\n❌ ERROR: {e}\")\n import traceback\n traceback.print_exc()\n\n # Emergency fallback submission\n print(\"\\nGenerating fallback submission (all zeros)...\")\n try:\n test_df = pd.read_csv(config.TEST_CSV)\n fallback = pd.DataFrame({\n 'sample_id': test_df['sample_id'],\n 'target': 0.0\n })\n fallback.to_csv(config.SUBMISSION_FILE, index=False)\n print(f\"✓ Fallback saved: {len(fallback)} rows\")\n except Exception as e2:\n print(f\"❌ Fallback failed: {e2}\")\n raise\n\n\nif __name__ == \"__main__\":\n run_inference()\nI always use the exact order (Green -> Dead -> Clover -> GDM -> Total), but I see your CSV is not in this order. Perhaps try this? This is the order from the overview of this comp, but I am not sure if this is the issue.
", "rawMarkdown": "I always use the exact order (Green -> Dead -> Clover -> GDM -> Total), but I see your CSV is not in this order. Perhaps try this? This is the order from the overview of this comp, but I am not sure if this is the issue.", "votes": null }, { "id": "3381033", "postDate": "12/23/2025 16:02:58", "content": "Since you did not include the full code I can't fork and find the error. As stated in many of the Submission Error posts, the first thing to do is to predict on the train.csv file rather than the test.csv.
\nMake that change - you will almost certainly get an error message when you run or Save/Run the code.
", "rawMarkdown": "Since you did not include the full code I can't fork and find the error. As stated in many of the Submission Error posts, the first thing to do is to predict on the train.csv file rather than the test.csv.\n\nMake that change - you will almost certainly get an error message when you run or Save/Run the code.", "votes": null }, { "id": "3381036", "postDate": "12/23/2025 16:07:19", "content": "I always have gdm last :) My bet is one of the two common errors. 1. Uses sample_submission.csv 2. Code only processes a single image
", "rawMarkdown": "I always have gdm last :) My bet is one of the two common errors. 1. Uses sample_submission.csv 2. Code only processes a single image", "votes": null }, { "id": "3381366", "postDate": "12/24/2025 10:46:49", "content": "I have tried, but it doesn't work. I don't know why my last correct submission can be activated. I only changed the model and then faced a format problem.
", "rawMarkdown": "I have tried, but it doesn't work. I don't know why my last correct submission can be activated. I only changed the model and then faced a format problem.", "votes": null }, { "id": "3381435", "postDate": "12/24/2025 15:36:38", "content": "Hmm - probably need to make the full model public for us to take a look at.
", "rawMarkdown": "Hmm - probably need to make the full model public for us to take a look at.", "votes": null }, { "id": "3381481", "postDate": "12/24/2025 17:53:12", "content": "How are you printing out the table in your post? Is it a dump of the actual submission.csv file that you've formatted by hand (or by other means) into a table, or is it a printout of a pandas dataframe with the same contents?
\nFor flushing out formatting stuff, I tend to rely on the output of this:
\nimport subprocess\nsubprocess.run([\"cat\",\"submission.csv\"])\nSometimes there are gotchas with pandas dataframe.to_csv(), like maybe it doesn't print out the header quite right if the dataframe and/or the to_csv() call aren't just so. In those cases, the output of cat can give you some insight into what's going wrong in the formatting.
I always use the exact order (Green -> Dead -> Clover -> GDM -> Total), but I see your CSV is not in this order. Perhaps try this? This is the order from the overview of this comp, but I am not sure if this is the issue.
", "votes": null, "replies": [ { "id": 3381036, "author_name": "pcjimmmy", "author_url": "", "post_date": "12/23/2025 16:07:19", "content": "I always have gdm last :) My bet is one of the two common errors. 1. Uses sample_submission.csv 2. Code only processes a single image
", "votes": null, "replies": [] }, { "id": 3381366, "author_name": "nguyenbaohuy05", "author_url": "", "post_date": "12/24/2025 10:46:49", "content": "I have tried, but it doesn't work. I don't know why my last correct submission can be activated. I only changed the model and then faced a format problem.
", "votes": null, "replies": [ { "id": 3381435, "author_name": "pcjimmmy", "author_url": "", "post_date": "12/24/2025 15:36:38", "content": "Hmm - probably need to make the full model public for us to take a look at.
", "votes": null, "replies": [] } ] } ] }, { "id": 3381033, "author_name": "pcjimmmy", "author_url": "", "post_date": "12/23/2025 16:02:58", "content": "Since you did not include the full code I can't fork and find the error. As stated in many of the Submission Error posts, the first thing to do is to predict on the train.csv file rather than the test.csv.
\nMake that change - you will almost certainly get an error message when you run or Save/Run the code.
", "votes": null, "replies": [] }, { "id": 3381481, "author_name": "particlebbq", "author_url": "", "post_date": "12/24/2025 17:53:12", "content": "How are you printing out the table in your post? Is it a dump of the actual submission.csv file that you've formatted by hand (or by other means) into a table, or is it a printout of a pandas dataframe with the same contents?
\nFor flushing out formatting stuff, I tend to rely on the output of this:
\nimport subprocess\nsubprocess.run([\"cat\",\"submission.csv\"])\nSometimes there are gotchas with pandas dataframe.to_csv(), like maybe it doesn't print out the header quite right if the dataframe and/or the to_csv() call aren't just so. In those cases, the output of cat can give you some insight into what's going wrong in the formatting.