IE 332 in-class session
Oct 29th: A2 & A3
¡ñ Find problems
Copyright By PowCoder代写 加微信 powcoder
¡ñ Solution for 4c should be
List the flight_IDs where more than one luggage can be checked using an economy class after October 2021.
A3 introduction & clarification
¡ñ Fibonacci Addition and Subtraction: See A3 tutorial 7
Supervised learning
¡ñ Use Spring 2020 As an example
¡ñ Steps of conducting supervised learning: See Tutorial 6
¡ñ What is problem with (b) and (c)?
Supervised learning
¡ñ Use Spring 2020 As an example
¡ñ Steps of conducting supervised learning: See Tutorial 6
¡ñ What is problem with (b) and (c)? Should split training and testing data before
scaling the data. Otherwise the test data would be scaled by its own center and scale, then the trained model would not be optimal for it.
¡ñ Which model is better? It depends on data and other methods you use(All models are wrong, but some are useful.
Supervised learning
¡ñ Use Spring 2020 As an example
¡ñ Start from Linear Regression
Supervised learning
¡ñ Use Spring 2020 As an example
¡ñ Start from Linear Regression
¡ñ Use Decision Tree
Supervised learning
¡ñ Use Spring 2020 As an example
¡ñ Start from Linear Regression
¡ñ Use Decision Tree
¡ñ Use Random Forest
Supervised learning
¡ñ Use Spring 2020 As an example
¡ñ Start from Linear Regression
¡ñ Use Decision Tree
¡ñ Use Random Forest
¡ñ Use SVM
¡ð Already need to tune many parameters
¡ð Observe the benefits of dimension reduction
Supervised learning
¡ñ Use Spring 2020 As an example
¡ñ Start from Linear Regression
¡ñ Use Decision Tree
¡ñ Use Random Forest
¡ñ Use SVM
¡ñ Use Historical Median
¡ñ Use Historical Median
Use Deep Learning
¡ñ Deep learning vs other machine learning methods: Professional camera vs iphone camera
¡ñ Requires much more experience: many hyperparameters to tune
¡ñ Need enough amount of data to work well
¡ñ Has a requirement on infrastructure to train them in reasonable time
¡ñ http://yann.lecun.com/exdb/mnist/ (Look for the best performance in nn and
svm, does it worth it, no defined answer?)
¡ñ really shines when it comes to complex problems such as image
classification, natural language processing, and speech recognition
RMSE Comparison
程序代写 CS代考 加微信: powcoder QQ: 1823890830 Email: powcoder@163.com