manisnesan/ml_notes

Scratch pad for studying ML

Coding

ML Basics

• Refer Machine Learning yearning book
• https://github.com/AstronomerAmber/ML_prep/blob/master/ML_phonescreen_Qs.md
• impl basic models Linear Reg, Logistic Reg, Naive Bayes, KNN[https://github.com/AstronomerAmber/ML_prep/blob/master/Practical_Coding_Challenges/kNN.md]
• Tradeoffs and Usecases [Refer https://github.com/sravya8/ML/blob/master/Other_Traditional_ML_Models.md]

ML Workflow

• Product Requirements

• Objective functions (Classification, Regression, Clustering, Segmentation)

• Datacollection and labeling

• EDA

• Evaluation Metrics

• Preprocess and Feature eng

• Split data into train, val, test

• Modeling with simple baseline

• Iterate based on bias/variance analysis

• Offline validation

• A/B testing ( business metrics != model/evaluation metrics)

• https://github.com/sravya8/ML/blob/master/Practical_ML.md

• https://manisnesan.github.io/chrestotes/intent-detection/text-classification/information-retrieval/query-understanding/machine-learning/2020/09/20/intent-detection.html

• https://github.com/sravya8/ML/blob/master/Practical_ML.md#improving-the-model-handling-bias-and-variance

ML UseCases

• Recommender Systems
• Search Ranking systems
• Classification

NN from scratch

• forward and backward passes
• initialization
• learning optimization
• conv layers
• rnn
• https://github.com/fastai/course22p2/blob/master/nbs/04_minibatch_training.ipynb

System Design

• Design a system providing a report on the number of views in Youtube
• DDIA