Generative-AI-with-AWS

types of machine learning

1. supervised learning: when you have labeled data, and want the model to learn the right answer from the labeled data
2. unsupervised learning: when you have massive dataset, unlabeled data, want the model to find structures and patterns in data.
3. reinforcement learning: when you want your model to learn like training a pet, interacting with the environment and receiving rewards/ penalties for its actions. Need feedback on the performance.
4. combination of multiple models and each model handle a specific task: e.g a self driving car integrate: one model using supervised learning to detect pedestrain, one model using unsupervised learning to cluster types of road condition, one model using reinforcement learning to make real time driving decision based on dynamic condition.

Train a RandomForestRegressor model to predict energy efficiency of the building based on the features of wall area, roof area, overall weight, glazing area. (supervised learning)

Use kmeans clustering to cluster vehicles based on their specifications (unsupervised learning)

Machine learning Approach

DecisionTreeClassifier to predict customer churn

A simple Neural Networks for predicting customer purchase behaviour based on 2 features: website visiting time and number of pages visited (binary classification problem)

1. generate synthetic data
2. preprocess the data: split the data
3. build and train the neural network
4. evaluate the model
5. visualization: for the training process we could visualiza the model accuracy and loss using the validation set visualize the decision boundary: to understand how the model differentiate between the 2 classes based on the website engagement metrics