# Notebooks - Canonical URL: https://leandeep.com/notebooks/ - Author: map[name:Olivier Eeckhoutte] - Published: 2014-04-10T00:00:00Z - Updated: 2014-04-10T00:00:00Z - Language: fr - License: CC BY-NC 4.0 (https://creativecommons.org/licenses/by-nc/4.0/) Last update of this page: May 2019 ## Notebooks et tips Here is a list of useful notebooks to handle different use cases: 1. [Gérer les gros datasets qui ne passent pas en mémoire](/datalab-own/huge-dataset.htm) 2. [Modèle complexe ou modèle simple ?](/datalab-own/modele-complexe-donnees-brutes-modele-simple-donnees-retravaillees.htm) 3. [Chemin le plus court](/datalab-own/chemin-le-plus-court.htm) 4. [Exemple d'analyse de données sur dataset déséquilibré](/datalab-own/analyse-donnees-dataset-desequilibre.htm) et [rapport d'analyse associé](/datalab-own/analyse-donnees-dataset-desequilibre_files/report.pdf) - Crédit: [Erdi Olmezogullari](https://github.com/erdiolmezogullari) 5. [Régression linéaire avec Tensorflow](/datalab-own/regression-lineaire-avec-tensorflow.html) 6. [Classification avec Tensorflow](/datalab-own/classification-avec-tensorflow.html) 7. [Underfitting and overfitting](/datalab-kaggle/underfitting-and-overfitting.html) **Spark recommandation engine example:** 1. [Recommandation part 1](/datalab-own/Spark/building-recommender.html) 2. [Recommandation part 2](/datalab-own/Spark/online-recommendations.html) 3. [Statistics mllib](/datalab-own/Spark/mllib-statistics.html)
## Compétitions Kaggle et kernels associés Here is a list of notebooks I studied and find interesting coming from Kaggle and having a MIT licence. I keep them here as inspirational sources or knowledge base. **Kaggle Winning solutions** 1. [Kaggle winning solutions](https://www.kaggle.com/sudalairajkumar/winning-solutions-of-kaggle-competitions) 2. [Kaggle winning solutions 2ème option](https://ndres.me/kaggle-past-solutions/) 3. [Kaggle winning solutions 3ème option](https://faridrashidi.github.io/kaggle-solutions/) **Kernels sur le sujet "titanic"** 1. [Good data exploration](/datalab-kaggle/good-data-exploration.html) 2. [Framework to achieve 99% accuracy](/datalab-kaggle/a-data-science-framework-to-achieve-99-accuracy.html) 3. [End to end Machine Learning pipeline](/datalab-kaggle/titanic-survival-prediction-end-to-end-ml-pipeline.html) 4. [Exploratory data analysis and feature extraction](/datalab-kaggle/data-analysis-and-feature-extraction-with-python.html) **Kernels sur le sujet "iris"** * [Full analysis](/datalab-own/Iris-Machine-Learning-Notebook.html) **Kernels autour du customer churn** * [Full analysis](/datalab-own/churn.html) **Kernels autour du jeu populaire battle royale** * [Fun EDA](/datalab-kaggle/eda-is-fun.html) **Kernels autour Google Analytics Customer Revenue Prediction** * [Simple Exploration+Baseline](/datalab-kaggle/simple-exploration-baseline-ga-customer-revenue.html) **Kernels autour des sales forecast** * [Sales Forecast ecommerce](/datalab-kaggle/e-commerce-sales-forecast.html) **Kernels sur le sujet "Porto Seguro’s Safe Driver Prediction"** * [Data preparation & exploration](/datalab-kaggle/data-preparation-exploration.html) **Recruit Restaurant Visitor Forecasting.** * [Top 25 solution](/datalab-kaggle/recruit-restaurant.html) **Sujets autour du sujet "House Pricing"** 1. [Full Workflow](/datalab-kaggle/DS-Workflow-for-HP-+10-ML-Algorithms-+12-Plot.html) 2. [regularized linear models](/datalab-kaggle/regularized-linear-models.html) 3. [Stacked Regressions](/datalab-kaggle/stacked-regressions-top-4-on-leaderboard.html) **Sujets autour de la compétition "home-credit-default-risk"** URL: https://www.kaggle.com/c/home-credit-default-risk To understand the data feel free to access [the "Data" chapter in this notebook](/datalab-kaggle/kb002.html#Data) of this notebook: Read-only Kernels: 1. [Basic end-to-end training of a LightGBM model](/datalab-kaggle/kb001.html) 2. Très bonnes explications sur la métrique ROC AUC et quelle métrique utiliser pour mesurer la performance d'un modèle lorsque des classes ne sont pas équilibrées ? Traite du _One Hot Encoding_ et du problème pour les features ayant trop de catégories (réduction de dimension). Explication sur le feature engineering et en particulier la construction de nouvelles features avec les méthodes _Polynomial Features_ et _Domain Knowledge Features_. [Home Credit Default Risk Competition - Full process - Part 1](/datalab-kaggle/kb002.html) 3. On continue à améliorer le modèle de la compétition home-credit-default-risk en créant de nouvelles features. On analyse si elles ont un impact sur la performance du modèle avec le coefficient de Pearson ou la Kernel Density Exploration (KDE). [Home Credit Default Risk Competition - Full process - Part 2](/datalab-kaggle/kb003.html) 4. (WIP) On continue encore à améliorer le modèle de la compétition home-credit-default-risk en utilisant toutes les données à disposition. [Home Credit Default Risk Competition - Full process - Part 3](/datalab-kaggle/kb004.html) **LUNA16 competition** * [Full preprocessing](/datalab-kaggle/luna-full-preprocessing-tutorial.html) **NYC taxi trip duration** * [Last cab to new york animated heatmap](/datalab-kaggle/last-cab-to-new-york-animated-heatmap-trips-folium.html) **Kernal autour du sujet Pokemon-Weedle's Cave** * [Full tutorial](/datalab-kaggle/pokemon-data-sciencetutorial-for-beginners.html) **Other kernels** 1. [Comprehensive data exploration](/datalab-kaggle/comprehensive-data-exploration-with-python.html) 2. [Handling missing values](/datalab-kaggle/handling-missing-values.html) 3. Fraud detection in Insurance - 3.1. [Part 1](/datalab-own/InsuranceClaimsFraudDetection/InsuranceClaimsFraudDetection.html) - 3.2. with Keras [Part 2](/datalab-own/InsuranceClaimsFraudDetection/InsuranceClaimsFraudDetection_UsingKeras.html)