Page dedicated to data exploratory analysis, preparation, cleaning, pre-processing / wrangling, generation, feature engineering and other related topics

The question to ask ourselves: Do we know our data...?

• Ethics / altruistic motives
• Data Science
• Datasets and sources of raw data
• Data Collection
• Hypothesis
• Data Exploratory Analysis
• Data preparation
  • Data Cleaning
  • Data preprocessing / Data Wrangling
• Data Generation
• Feature Extraction
• Feature Importance
• Feature Engineering
• Feature Selection
• Hyperparameter tuning
• Post model-creation analysis, ML interpretation/explainability
• Model deployment
• Statistics
• Visualisation
• Common mistakes when training models (data related)
• Presentations
• Cheatsheets
• Course / books
• Best practices / rules / an unordered list of high level or low level guidelines
• Framework(s) / checklist(s)
• Notebooks
• Programs and Tools
• Databases
• References
• Credits
• Contributing

See Ethics / altruistic motives

• The Data Science Process
  • Frame The Problem
  • Usage and Integration
  • Starting a Data Project
  • WorkSpace Setup and Cloud Computing
• JustCause package/framework - framework to foster good scientific practice in the research of causality methods | PyPu | GitHub
• “Metaflow is a human-friendly Python library” LinkedIn Post
• 5 free books for learning Python for DS
• 7 advanced tricks in pandas for data science
• The Ultimate NumPy Tutorial for Data Science Beginners
• Top 10 Data science podcast must follow for learn new things
• Top 20 Youtube Channels for Data Science
• Advanced Data Science from IBM
• 𝟏𝟐 𝐒𝐭𝐞𝐩𝐬 𝐭𝐨 𝐏𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐨𝐧-𝐐𝐮𝐚𝐥𝐢𝐭𝐲 𝐃𝐚𝐭𝐚 𝐒𝐜𝐢𝐞𝐧𝐜𝐞 𝐂𝐨𝐝𝐞
• Top 10 Popular GitHub Repositories to learn about Data Science
• The difference between Statistics and Data Science: Big Data and Inferential Statistics
• DataScience resources (in the form of a book) from Eric
• Data Exploration and API First Design: Deep Learning Hands-On Series with Eric Schles
• Augmented Analytics Engine
• Putting an end to Unreliable Analytics by David Yaffe
• The Fundamentals of end-to-end Data Strategy: video | slides | Resources | Feedback

See Datasets

• 👉 Effective Data Collection 👈
• The Ultimate Guide to Effective Data Collection

• Correlation, causation, multicollinearity, confounding features or variables
• How to approach Hypothesis Testing
• Does Your Hypothesis Development Canvas Tell a Story?
• A Complete Guide to Hypothesis Testing
• An introduction to Statistical Inference and Hypothesis testing
• A set of descriptive statistics and hypothesis tests across different types of data
• The statistical analysis t-test explained for beginners and experts
• The book of why" by Judea Pearl : A great overview and presents many relevant techniques
• [Craig's presentation: Visualizing the Why — Strategy and Roadmaps in Context](https://www.dropbox.com/s/knagl9f7u9hxvr7/Strategy%20Maps%20Agile%20Evangelists.pdf?dl=0](https://www.youtube.com/watch?v=LOjsuYzzOkA)
• Correlation & Causation: The Couple That Wasn’t

See Data Exploratory Analysis

• Data preparation
• See Data preparation

See Data Generation

• Hierarchical Feature Extraction for Compact Representation and Classification of Datasets
• Guide to Feature Extraction Approaches for Text Data

• Example: Feature Importance implementation (python)
• How to Calculate Feature Importance With Python
• RFPimp:
  • RF Importance
  • Explaining Feature Importance by example of a Random Forest
• Catboost model and W&B
• LightGBM model and W&B
• The 4 types of additive Feature Importances
• The Math of Random Forests and Feature Importance in Scikit-learn and Spark
• Path Explain - toolkit for feature attributions: GitHub | PyPI | Path Explain on MWML
• Open Machine Learning Course: Feature Importance

See Feature engineering

See Feature Selection

• Ray Tune Sweeps
• W&B Sweeps
• Automated Machine Learning Hyperparameter Tuning in Python
• Bayesian hyperparameter optimisation by Akinkunle: Original Notebook | Saved Notebook | Slides
• Hyperparameter optimization for Neural Networks
• Tune Hyperparameters Easily with W&B Sweeps

• Pruning: DL models
• [Pruning models](https://app.wandb.ai/authors/pruning/reports/Plunging-into-Model-Pruning-in-Deep-Learning--VmlldzoxMzcyMDg](https://app.wandb.ai/authors/pruning/reports/Scooping-into-Model-Pruning-in-Deep-Learning--VmlldzoxMzcyMDg?utm_source=social_slack&utm_medium=slack&utm_campaign=report_author)
• Poor Man’s BERT • Exploring Pruning as an Alternative to Knowledge Distillation. See Post model-creation analysis, ML interpretation/explainability

• Model Deployment Methods and Techniques - Part 1
• Model Deployment Methods and Techniques - Part 2
• Model Deployment Methods and Techniques - Part 3
• Model Deployment Methods and Techniques - Part 4
• Model Deployment Methods and Techniques - Part 5

• Mode of a Log-Normal distribution by Sahil Gupta See Statistics.md

• Data Visualization
• See Visualisation

• Having a lot more training examples of one type of object than the other types
• Accidentally testing the neural network using images that were in the training set
• Training the neural network on data that is easier to recognize or more consistent than the real-world data it will be used to classify later on

• A Rubric for ML Production Readiness - by Jiameng Gao from Applied Deep Learning Meetup in Feb 2019 (Paper: https://ai.google/research/pubs/pub46555)
• Introduction to Data Analysis and Cleaning presentation by Mark Bell
• Do we know our data, as good as we know our tools by Jeremie Charlet and Mani Sarkar

See under Cheatsheets

See Courses / books

• 12 Best Practices for Modern Data Ingestion
  • PDF
• A Rubric for ML Production Readiness - by Jiameng Gao from Applied Deep Learning Meetup in Feb 2019 (Paper: https://ai.google/research/pubs/pub46555)
• Rules of Machine Learning: Best Practices for ML Engineering

See Framework(s) / checklist(s)

See Notebooks

See Programs and Tools

See Databases

• How to build a data science project from scratch
• Common mistakes when carrying out machine learning and data science
• A Rubric for ML Production Readiness - Breck et al. 2017 by Jiameng Gao (28 rules to follow, suggested by Google) | Original Paper by Google
• Understanding Data Science Problems - template of questions to ask
• eBook: How to Succeed in Data Science [deadlink]
• Data Fallacies by Nabih Bawazir

Big thanks to Jeremie Charlet for his contributions to many of the resources on this page. Not forgetting the others who have also helped support in the building of the above resources.

Contributions are very welcome, please share back with the wider community (and get credited for it)!

Please have a look at the CONTRIBUTING guidelines, also have a read about our licensing policy.

Back to main page (table of contents)