H2O and R: Need help with running Airline dataset using GLM model

[Anu Bulusu]

Hello,

I am a new user of H2O and R. I am trying to use the airline dataset to run a GLM model using H2O in R (without regularization). I am encountering numerous errors in my code as shown below in red. Can someone please help. - Thanks for your time.

library(h2o)

localH2O = h2o.init(nthreads = -1)  # Initializes one node in the H2O

pathToData = normalizePath("airline.csv")

# Loading the data from the local dir

airline.hex = h2o.uploadFile(path = pathToData, header = TRUE, destination_frame = "airline.hex")

# Splitting the dataset into training and validation objects

# Method 1

s = h2o.runif(airline.hex, seed = 1234)

airline.train = airline.hex[s <= 0.8,]

airline.test = airline.hex[s > 0.8,]

# Method 2

#airline.split = h2o.splitFrame(data = airline.hex, ratios = 0.8)

#airline.train = airline.split[[1]]

#airline.test = airline.split[[2]]

summary(airline.train)

summary(airline.test)

x = c("Year", "Month", "DayofMonth", "DayOfWeek", "UniqueCarrier", "Origin", "Dest", "Distance")

y = "IsDepDelayed"

# Training GLM model

airline.model <- h2o.glm(x=x, y=y, training_frame = airline.train, validation_frame = airline.test, model_id = "glm.model", family = "binomial", lambda_search = FALSE)

# Extracting and handling results

#(1) Making predictions

print("Predict on GLM model")

airline.results = h2o.predict(airline.model, airline.test)

#(2) Calculating metrics Performance and AUC on results file

print("Check performance and AUC")

perf = h2o.performance(airline.model, airline.results)

auc  = h2o.auc(perf)

#Error in .h2o.doSafeREST(conn = conn, h2oRestApiVersion = h2oRestApiVersion,  : Test/Validation dataset has no columns in common with the training set

# If I comment out the perf and aux metric commands above and run, I get the next error listed below

print("Show Distribution of predictions with quantile")

quant = h2o.quantile(airline.results[,2])

print("Extract strongest predictions")

top.airline <- h2o.assign(airline.results[airline.results$YES > quant["75%"]], key = "top.airline")

#Error in expr[[1L]] : subscript out of bounds

top.airline

# Perform classification on validation frame

prediction = h2o.predict(object = airline.model, newdata = airline.test)

# Copy predictions from H2O to R

pred = as.data.frame(prediction)

head(pred)

[Amy Wang]

Hey Anu,

So the documentation ang vignette for H2O-3 is actually available here: http://h2o.ai/resources/

The vignette you found at leanpub is deprecated and for version h2o-classic.

I've also attached a script fixes up all the arguments and inputs.

Sincerely,

Amy Wang