h2o R Example

h2o_example.r

# Load h2o
library(h2o)
library(ggplot2)

# Load Dataset - London Bike 
london_bike <- read.csv('./london_merged.csv')

# Transforming Weather code and Season to factor
london_bike$weather_code <- as.factor(london_bike$weather_code)
london_bike$season <- as.factor(london_bike$season)

h2o.init()

# Convert london_bike to h2o frame
london_bike.h2o <- as.h2o(london_bike)

# Split london_bike into train_test split
london_bike_split <- h2o.splitFrame(data = london_bike.h2o, ratios = 0.8, seed = 1234)

training_data <- london_bike_split[[1]]
test_data <- london_bike_split[[2]]

# Training Linear regression

predictors <- c("t1", "t2", "hum", "wind_speed", "weather_code", "is_holiday",
                "is_weekend", "season")
response <- "cnt"


# Model 1
london_bike_model <- h2o.glm(x = predictors,
                      y = response,
                      training_frame = training_data)

test_predict <- h2o.predict(object = london_bike_model, 
                            newdata = test_data)

predictions_x_real <- cbind(
  as.data.frame(test_data$cnt),
  as.data.frame(test_predict)
)

ggplot(
  data = as.data.frame(predictions_x_real),
  aes(x=cnt, y=predict)
) + geom_point(color='darkgreen') + xlab('Actual Label') + ylab('Predictions')

# Training Linear Regression using Regularization
london_bike_model_regularized <- h2o.glm(x = predictors,
                             y = response,
                             training_frame = training_data,
                             alpha = 1)

test_predict_regularized <- h2o.predict(object = london_bike_model_regularized, 
                            newdata = test_data)

predictions_x_real_regularized <- cbind(
  as.data.frame(test_data$cnt),
  as.data.frame(test_predict_regularized)
)

ggplot(
  data = as.data.frame(predictions_x_real_regularized),
  aes(x=cnt, y=predict)
) + geom_point(color='darkgreen') + xlab('Actual Label') + ylab('Predictions')

# Evaluating models - Using validation_frame
london_bike_model <- h2o.glm(x = predictors,
                             y = response,
                             training_frame = training_data,
                             validation_frame = test_data)

# 
h2o.rmse(london_bike_model, train=TRUE, valid=TRUE)

# Random Forest Example
london_bike_rf <- h2o.randomForest(x = predictors,
                                   y = response,
                                   ntrees = 25,
                                   max_depth = 5,
                                   training_frame = training_data,
                                   validation_frame = test_data)

# Retrieving metrics for randomforest
h2o.rmse(london_bike_rf, train=TRUE, valid=TRUE)
h2o.r2(london_bike_rf, train=TRUE, valid=TRUE)

# Training Neural Network
nn_model <- h2o.deeplearning(x = predictors,
                             y = response,
                             hidden = c(6,6,4,7),
                             epochs = 1000,
                             train_samples_per_iteration = -1,
                             reproducible = TRUE,
                             activation = "Rectifier",
                             seed = 23123,
                             training_frame = training_data,
                             validation_frame = test_data)

# Neural Network Evaluation
h2o.rmse(nn_model, train=TRUE, valid=TRUE)
h2o.r2(nn_model, train=TRUE, valid=TRUE)

# Grid Search 
rf_params <- list(ntrees = c(2, 5, 10, 15),
                    max_depth = c(3, 5, 9),
                    min_rows = c(5, 10, 100))

# Train and validate a grid of randomForests
rf_grid <- h2o.grid("randomForest", 
                      x = predictors, 
                      y = response,
                      grid_id = "rf_grid",
                      training_frame = training_data,
                      validation_frame = test_data,
                      seed = 1,
                      hyper_params = rf_params)


h2o.getGrid(grid_id = "rf_grid",
            sort_by = "r2",
            decreasing = TRUE)

# Auto ML Routine
aml <- h2o.automl(x = predictors, 
                  y = response,
                  training_frame = training_data,
                  validation_frame = test_data,
                  max_models = 15,
                  seed = 1)
                  

# Explainability
london_bike_rf <- h2o.randomForest(x = predictors,
                                   y = response,
                                   ntrees = 25,
                                   max_depth = 5,
                                   training_frame = training_data,
                                   validation_frame = test_data)



# Variable importance plot
h2o.varimp_plot(london_bike_rf)

# Shap Summary
h2o.shap_summary_plot(london_bike_rf, test_data)

# Shap Explain Row
h2o.shap_explain_row_plot(london_bike_rf, test_data, row_index = 4)

# Shap Explain Summer Row
h2o.shap_explain_row_plot(london_bike_rf, test_data, row_index = 830)