Please enable JavaScript.
Coggle requires JavaScript to display documents.
Clarifying Requirements - Coggle Diagram
Clarifying Requirements
Framing the Problem as an ML Task
Data Preparation
Model Development
Evaluation
Development and Serving
Monitoring and Infrastructure
System Failure
Data Distribution Shift Solutions
Train on large datasets
Regularly retrain
Data Augmentation?
What to monitor
Operation-related metrics
ML-spcific metrics
Infrastructure
Cloud v.s. On-device
Model compression
Knowledge Distillation
Pruning
Quantization
Testing in Production
Shadow Deployment
A/B Testing
Prediction Pipeline
Batch Prediction
Online Prediction
Offline Evaluation
Online Evaluation
Model Selection
Logistic regression
Linear Regression
DT
Random Forestes
SVM
Naive Bayes
Neural networks
Model Training
Construct Dataset
Choose Loss Function
Training
From Scratch
Fine-Tuning
Distributed Training
Collect Data
Identify Features and Labels
Proper Sampling
1 more item...
Data Engineering
Feature Engineering
Select Features(with domain knowledge)
Format Transformation
Handle missing values
Delete
Imputate
Scale
Min-Max
Standardization
Log
Bucketing
Encoding
Interger encoding
one-hot encoding
Embedding Learning
Data Source
Data Storage
Data Type
structured
Numerical
Discrete
Continuous
Categorical
Ordinal
Nominal
Unstructured
Audio
Video
Image
Text
Define ML objective
Specify input and output
Choose the ML category
Supervised Learning
Classification
Regression
Unsupervised Learning
Reinforcement Learning