IEML

Introduction to Exploratory Machine Learning

• Experiments with Muller Guido book
• Bishop, Pattern Recognition and Machine Learning

0. Welcome Party

1. Environment Setup

• Python 3.7
• VSCode (Free Open Source)
• Jupyter-lab
• Python Programming Revisited.
  • www.jungol.co.kr

2. Basic Python libraries

• PIL: Python Imaging library (pip3 install pillow)
• Numerical Computation: numpy
• Visualization: matplotlib.pyplot
• Several simple problem solving with numpy
  • random sampling functions
  • sampling from probability density models

3. Python libraries II

• pandas for tabular data maipulation as a replacement of Excel/Spreedsheet
• Scientific computing: scipy & scikit-learn

4. Probability and Statistics, revisted

• ref: chapter 2 of PRML (Pattern Recognition ad Machine Learning by Christopher M. Bishop)
• various probability density models
• problem solving

5. Bayes' Theorem

• Intro. to Bayesian Data Analysis(https://www.youtube.com/watch?v=3OJEae7Qb_o)

6. Bayesian Data Analysis, short intro & problem solving

• Think Bayes

7. Supervisded Learning: Regression

8. Well-being Party

9. Supervised Learning: Classification

10. Unsupervised Learning

• Dimensionality Reduction, Feature Extraction, Manifold Learning
  • PCA, NMF, t-SNE

11. Unsupervised Learning

• Clustering
  • k-Means, Agglomerative Clutstering, DBSCAN

12. Feature Engineering

• common sense (New Yong Citi Bike)

13. Model Evaluation and Improvement

• Cross validation
• Grid Search
• Accuracy, Precision, Recall, F1

14. Algorithm Chains and Pipelines

15. Project Show-up

16. Farewell Party

Data

• Datasets

  • IRIS
  • MNIST
  • Seoul City Public Bike Rental Data (https://data.seoul.go.kr/dataList/datasetView.do?infId=OA-15182&serviceKind=1&srvType=F)
    • Seoul City Bike Info from data.go.kr: https://www.data.go.kr/dataset/3045310/fileData.do

• Kaggle

  • https://www.kaggle.com/c/bike-sharing-demand/overview
  • https://www.kaggle.com/c/titanic
  • https://www.kaggle.com/c/digit-recognizer

Still under development