Skip to contents
ClassRegPred
1.0
Part 1
Running regressions is easy ....
LPM and The Titanic data: Survived or not ?
LPM and The Titanic data: Survived or not ?
Discrete response models: The Generalised Linear Models (GLM)
GLM / Logit and The Titanic data: Survived or not ?
GLM / Logit and The Titanic data: Survived or not ?
GLM / Probit: Female labour supply
GLM / Probit: Female labour supply
GLM / Probit / Logit: Discrimination on the labour market
GLM / Poisson Regressions: The Bikeshare Data
GLM / Poisson Regressions: The Bikeshare Data
Resampling methods: Cross-validation (CV)
Regression trees
Classification trees
Random Forests: The Titanic data (Survived or not) ?
Random Forests: The Titanic data (Survived or not) ?
Comparison of predictive performance of classification methods: The Titanic data
Comparison of predictive performance of classification methods: The Titanic data
Regularisation
Regularisation
Part 2 (DL)
Neural Networks
Gradient Descent
NN Function Approximations
NN preliminaries (py torch)
simple NN (hitters data)
multilayer NN (MNIST)
CNN 1: CIFAR-1
CNN 2: deploy post training
cNN training: RESNET and CIFAR-100 (gpu, colab)
cNN post-training deployment (gpu, colab)
GPT
NLP tasks
Occam's razor: OLS v. Deep Learning v. Lasso
Occam's razor: OLS v. Deep Learning v. Lasso
Gallery
coding
Homework
HW1: OLS and the boat views data
HW2: Data wrangling using the (raw) boat views data
HW3: Modelling excess mortality in France to predict covid death
HW4: Logit model: Spam or not in the Email data set?
HW5: Cross-validation (CV): Spam and Logit
HW6: Random Forests: Spam or not in the Email data set?
Extras
Occam's razor: OLS v. Deep Learning v. Lasso
Occam's razor: OLS v. Deep Learning v. Lasso
Streamlining the Work Flow: tidymodels (parsnip and recipes)
Streamlining the Work Flow: Pipelines (OLS and Lasso using Ames data)
Automated feature selection: Recursive feature elimination (RFE)
Complete projects
Random Forests in caret: Predicting probabilities
Predicting whether an email is spam in the Email Data
Predicting whether an email is spam in the Email Data
Predicting whether a flight is delayed in the NYC flights data
Predicting whether a flight is delayed in the NYC flights data
Random Forests in caret: Predicting probabilities for the flight data
Predicting sex using historical data for first names (USA-IPUMS)
CV and performance metrics for house price predictions
Predicting boat views
Predicting boat views
Predicting high site visits
Predicting high site visits
Multiclass prediction using the Dry-Beans data
projects
Predicting high site traffic
Light
Dark
Auto
Gallery: ClassRegPred
gallery_crp.Rmd
Methods
Methods: LPM, GLM, RFs (Titanic)
OLS predictions: Excess covid deaths
Neural Networks
Gradient Descent
shallow NN
shallow NN: universal approx. theorem
CNNs: Minst data
NLP (GPT,T5,BART,…): Token classification (NER), seq2seq, etc.
RL
Deep RL: Gridworld
Deep RL: Lunar Lander
Deep RL: Space Invaders
Data sets
Survival on the Titanic
Boat views
Site visits
CNNs: Minst data
CNNs: CIFAR100