Please enable JavaScript.
Coggle requires JavaScript to display documents.
Data Modeling - Coggle Diagram
Data Modeling
Overview of Modeling
Neural Networks
- Made up of neurons (nodes)
- One neuron is responsible for processing inputs and producing an output
- A series of computational layers
Activation functions
- Functions we can use as decision boundaries to tell us whether or not to send a signal to subsequent layers in the network
- Allows us to pass a range of values to the following layer
- Important for data flow in the model
-
-
Purpose Built Models
Multi-Layer Perceptron
- Basic perceptron model
- Simple data types
Convolutional Neural Networks
- Convolutional & pooling
- Common in computer vision
Long Short Term Memory (LSTM) Network
- Can save internal state
- Common in NLP
-
Model Evaluation
Understanding Model Performance
- Need robust metrics to know how model will perform after deployment
- Knowing which metrics to measure will help guide model development
- Model performance will determine overall success of AI products
Labeled data
- 80% training data
- 10% validation data
- 10% test data
- A model should never see the test data until model evaluation
- Precision & recall are the key metrics when evaluating a model
- F1 score provides an overall measure of model performance
- A confusion matrix can help identify where a model is failing
-