{
  "id": 327924,
  "title": "Collection of discussions for Amex Competition",
  "url": "/competitions/amex-default-prediction/discussion/327924",
  "author_name": "X_T_X",
  "post_date": "2022-05-30T02:52:46.215000",
  "votes": 9,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi, everyone. Just a collection of discussion topics for this competition organized in one place. Hope this can be helpful. WIP.</p>\n<p><strong>Past Insights:</strong><br>\nArticles, Research Papers and Methodologies:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327135\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327135</a></p>\n<p>Insights from a previous default prediction competition<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327148\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327148</a></p>\n<p>Let's catchup with all the learnings so far<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328565\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328565</a></p>\n<p>An obvious pointer…<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327922\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327922</a></p>\n<p><strong>Datesets:</strong><br>\nTraining data starter:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327106\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327106</a></p>\n<p>Tutorial on reading large datasets by Rohan<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327205\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327205</a></p>\n<p>Reading &amp; Working with Large Dataset:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327195\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327195</a></p>\n<p>6.53GB Dataset + Kaggle Notebook training and Inference Pipeline:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327228\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327228</a></p>\n<p>Handling large datasets with Dask<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327110\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327110</a></p>\n<p>Compressed Dataset with targets (~10x compression) &amp; Notebook:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327268\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327268</a></p>\n<p>How to reduce pandas memory while loading dataframe ?<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327333\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327333</a></p>\n<p>[FAST LOADING] only 1.4GB Training Data using Feather<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327400\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327400</a></p>\n<p>10x Compression of whole dataset using Feather<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327441\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327441</a></p>\n<p>Last month per customer<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327094\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327094</a></p>\n<p>Another way to keep only the last month without groupby<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327361\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327361</a></p>\n<p>Kaggle Dataset for Transformers and RNNs<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327828\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327828</a></p>\n<p>Parquet Format Dataset for Low Memory Use<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327138\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327138</a></p>\n<p>⚡ 9x Data Compression achieved with Feather<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327143\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327143</a></p>\n<p>Segmentation Fault 😱 with PyArrow &amp; Dask<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327965\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327965</a></p>\n<p>How To Reduce Data Size<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328054\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328054</a></p>\n<p>Risk of overflow with Float 16 conversion<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328138\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328138</a></p>\n<p>Integer columns in the data - here you go!<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328514\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328514</a></p>\n<p><strong>Features/target related:</strong><br>\nUnique number of categorical features:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327161\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327161</a></p>\n<p>Default Definition<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327158\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327158</a></p>\n<p>About the nature of the target<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327367\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327367</a></p>\n<p>Strange Histograms<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327651\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327651</a></p>\n<p>Time Series **EDA ** and GRU Starter - LB 0.790<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327761\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327761</a></p>\n<p>Advanced **EDA **- UMAP/Hdbscan<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328084\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328084</a></p>\n<p>Minority Report<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327597\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327597</a></p>\n<p>The data has uniform random noise injected<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327649\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327649</a></p>\n<p>Analysis of Information loss during conversion from float64 to float16<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328057\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328057</a></p>\n<p><strong>Evaluation metric:</strong></p>\n<p>What does it mean?<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327234\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327234</a></p>\n<p>Amex metric using pd.Series<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327162\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327162</a></p>\n<p>Metric without DF<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327534\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327534</a></p>\n<p>Custom metric without pandas DataFrame, and accelerated by numba<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327609\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327609</a></p>\n<p>Graphical explanation of the competition metric<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327464\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327464</a></p>\n<p>Normalized Gini Coefficient (G). Default Rate (D).<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327116\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327116</a></p>\n<p>Evaluation Metric in Datatable (3x speedup over Pandas)<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327984\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327984</a></p>\n<p>10x fast metric (numpy)<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328020\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328020</a></p>\n<p><strong>Models:</strong><br>\nGBDT or NN,which is the winner of this competition<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327765\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327765</a></p>\n<p>Hopular with GBDT ….so XGBoost is all you need =)))<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328801\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328801</a></p>\n<p>Speed Up XGB, CatBoost, and LGBM by 20x<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328606\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328606</a></p>",
  "messages": [
    {
      "id": 1805267,
      "postDate": "2022-05-30T02:52:46.217Z",
      "content": "<p>Hi, everyone. Just a collection of discussion topics for this competition organized in one place. Hope this can be helpful. WIP.</p>\n<p><strong>Past Insights:</strong><br>\nArticles, Research Papers and Methodologies:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327135\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327135</a></p>\n<p>Insights from a previous default prediction competition<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327148\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327148</a></p>\n<p>Let's catchup with all the learnings so far<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328565\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328565</a></p>\n<p>An obvious pointer…<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327922\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327922</a></p>\n<p><strong>Datesets:</strong><br>\nTraining data starter:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327106\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327106</a></p>\n<p>Tutorial on reading large datasets by Rohan<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327205\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327205</a></p>\n<p>Reading &amp; Working with Large Dataset:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327195\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327195</a></p>\n<p>6.53GB Dataset + Kaggle Notebook training and Inference Pipeline:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327228\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327228</a></p>\n<p>Handling large datasets with Dask<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327110\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327110</a></p>\n<p>Compressed Dataset with targets (~10x compression) &amp; Notebook:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327268\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327268</a></p>\n<p>How to reduce pandas memory while loading dataframe ?<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327333\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327333</a></p>\n<p>[FAST LOADING] only 1.4GB Training Data using Feather<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327400\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327400</a></p>\n<p>10x Compression of whole dataset using Feather<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327441\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327441</a></p>\n<p>Last month per customer<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327094\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327094</a></p>\n<p>Another way to keep only the last month without groupby<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327361\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327361</a></p>\n<p>Kaggle Dataset for Transformers and RNNs<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327828\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327828</a></p>\n<p>Parquet Format Dataset for Low Memory Use<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327138\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327138</a></p>\n<p>⚡ 9x Data Compression achieved with Feather<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327143\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327143</a></p>\n<p>Segmentation Fault 😱 with PyArrow &amp; Dask<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327965\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327965</a></p>\n<p>How To Reduce Data Size<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328054\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328054</a></p>\n<p>Risk of overflow with Float 16 conversion<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328138\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328138</a></p>\n<p>Integer columns in the data - here you go!<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328514\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328514</a></p>\n<p><strong>Features/target related:</strong><br>\nUnique number of categorical features:<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327161\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327161</a></p>\n<p>Default Definition<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327158\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327158</a></p>\n<p>About the nature of the target<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327367\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327367</a></p>\n<p>Strange Histograms<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327651\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327651</a></p>\n<p>Time Series **EDA ** and GRU Starter - LB 0.790<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327761\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327761</a></p>\n<p>Advanced **EDA **- UMAP/Hdbscan<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328084\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328084</a></p>\n<p>Minority Report<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327597\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327597</a></p>\n<p>The data has uniform random noise injected<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327649\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327649</a></p>\n<p>Analysis of Information loss during conversion from float64 to float16<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328057\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328057</a></p>\n<p><strong>Evaluation metric:</strong></p>\n<p>What does it mean?<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327234\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327234</a></p>\n<p>Amex metric using pd.Series<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327162\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327162</a></p>\n<p>Metric without DF<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327534\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327534</a></p>\n<p>Custom metric without pandas DataFrame, and accelerated by numba<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327609\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327609</a></p>\n<p>Graphical explanation of the competition metric<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327464\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327464</a></p>\n<p>Normalized Gini Coefficient (G). Default Rate (D).<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327116\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327116</a></p>\n<p>Evaluation Metric in Datatable (3x speedup over Pandas)<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327984\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327984</a></p>\n<p>10x fast metric (numpy)<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328020\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328020</a></p>\n<p><strong>Models:</strong><br>\nGBDT or NN,which is the winner of this competition<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/327765\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/327765</a></p>\n<p>Hopular with GBDT ….so XGBoost is all you need =)))<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328801\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328801</a></p>\n<p>Speed Up XGB, CatBoost, and LGBM by 20x<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/328606\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/328606</a></p>",
      "rawMarkdown": "Hi, everyone. Just a collection of discussion topics for this competition organized in one place. Hope this can be helpful. WIP.\n\n**Past Insights:**\nArticles, Research Papers and Methodologies:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327135\n\nInsights from a previous default prediction competition\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327148\n\nLet's catchup with all the learnings so far\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328565\n\nAn obvious pointer...\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327922\n\n**Datesets:**\nTraining data starter:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327106\n\nTutorial on reading large datasets by Rohan\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327205\n\nReading & Working with Large Dataset:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327195\n\n6.53GB Dataset + Kaggle Notebook training and Inference Pipeline:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327228\n\nHandling large datasets with Dask\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327110\n\nCompressed Dataset with targets (~10x compression) & Notebook:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327268\n\nHow to reduce pandas memory while loading dataframe ?\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327333\n\n[FAST LOADING] only 1.4GB Training Data using Feather\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327400\n\n10x Compression of whole dataset using Feather\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327441\n\nLast month per customer\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327094\n\nAnother way to keep only the last month without groupby\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327361\n\nKaggle Dataset for Transformers and RNNs\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327828\n\nParquet Format Dataset for Low Memory Use\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327138\n\n⚡ 9x Data Compression achieved with Feather\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327143\n\nSegmentation Fault 😱 with PyArrow & Dask\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327965\n\nHow To Reduce Data Size\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328054\n\nRisk of overflow with Float 16 conversion\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328138\n\nInteger columns in the data - here you go!\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328514\n\n\n**Features/target related:**\nUnique number of categorical features:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327161\n\nDefault Definition\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327158\n\nAbout the nature of the target\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327367\n\nStrange Histograms\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327651\n\nTime Series **EDA ** and GRU Starter - LB 0.790\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327761\n\nAdvanced **EDA **- UMAP/Hdbscan\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328084\n\n\nMinority Report\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327597\n\nThe data has uniform random noise injected\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327649\n\nAnalysis of Information loss during conversion from float64 to float16\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328057\n\n**Evaluation metric:**\n\nWhat does it mean?\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327234\n\nAmex metric using pd.Series\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327162\n\nMetric without DF\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327534\n\nCustom metric without pandas DataFrame, and accelerated by numba\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327609\n\nGraphical explanation of the competition metric\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327464\n\nNormalized Gini Coefficient (G). Default Rate (D).\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327116\n\nEvaluation Metric in Datatable (3x speedup over Pandas)\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327984\n\n10x fast metric (numpy)\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328020\n\n\n**Models:**\nGBDT or NN,which is the winner of this competition\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327765\n\nHopular with GBDT ....so XGBoost is all you need =)))\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328801\n\nSpeed Up XGB, CatBoost, and LGBM by 20x\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328606\n\n\n",
      "votes": 9
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1805267": "Hi, everyone. Just a collection of discussion topics for this competition organized in one place. Hope this can be helpful. WIP.\n\n**Past Insights:**\nArticles, Research Papers and Methodologies:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327135\n\nInsights from a previous default prediction competition\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327148\n\nLet's catchup with all the learnings so far\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328565\n\nAn obvious pointer...\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327922\n\n**Datesets:**\nTraining data starter:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327106\n\nTutorial on reading large datasets by Rohan\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327205\n\nReading & Working with Large Dataset:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327195\n\n6.53GB Dataset + Kaggle Notebook training and Inference Pipeline:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327228\n\nHandling large datasets with Dask\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327110\n\nCompressed Dataset with targets (~10x compression) & Notebook:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327268\n\nHow to reduce pandas memory while loading dataframe ?\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327333\n\n[FAST LOADING] only 1.4GB Training Data using Feather\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327400\n\n10x Compression of whole dataset using Feather\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327441\n\nLast month per customer\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327094\n\nAnother way to keep only the last month without groupby\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327361\n\nKaggle Dataset for Transformers and RNNs\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327828\n\nParquet Format Dataset for Low Memory Use\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327138\n\n⚡ 9x Data Compression achieved with Feather\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327143\n\nSegmentation Fault 😱 with PyArrow & Dask\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327965\n\nHow To Reduce Data Size\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328054\n\nRisk of overflow with Float 16 conversion\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328138\n\nInteger columns in the data - here you go!\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328514\n\n\n**Features/target related:**\nUnique number of categorical features:\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327161\n\nDefault Definition\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327158\n\nAbout the nature of the target\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327367\n\nStrange Histograms\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327651\n\nTime Series **EDA ** and GRU Starter - LB 0.790\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327761\n\nAdvanced **EDA **- UMAP/Hdbscan\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328084\n\n\nMinority Report\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327597\n\nThe data has uniform random noise injected\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327649\n\nAnalysis of Information loss during conversion from float64 to float16\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328057\n\n**Evaluation metric:**\n\nWhat does it mean?\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327234\n\nAmex metric using pd.Series\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327162\n\nMetric without DF\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327534\n\nCustom metric without pandas DataFrame, and accelerated by numba\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327609\n\nGraphical explanation of the competition metric\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327464\n\nNormalized Gini Coefficient (G). Default Rate (D).\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327116\n\nEvaluation Metric in Datatable (3x speedup over Pandas)\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327984\n\n10x fast metric (numpy)\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328020\n\n\n**Models:**\nGBDT or NN,which is the winner of this competition\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/327765\n\nHopular with GBDT ....so XGBoost is all you need =)))\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328801\n\nSpeed Up XGB, CatBoost, and LGBM by 20x\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/328606\n\n\n"
  }
}