generateMCPrediction: generateMCPrediction

Description

Predict results on test data using the tuned MarkovChain mode. Result has the required data format for a submission to the Kaggle COVID-19 challenge.

Usage

generateMCPrediction(
  testData,
  models,
  startSimulation = "2020-01-22",
  write = FALSE
)

Arguments

testData

'data.frame': obs. of 3 variables:

ForecastId: int 1 2 3 4 5 6 7 8 9 10 ...
Region: Factor w/ 313 levels "Afghanistan/",..: 1 1 1 1 1 1 1 1 1 1 ...
Date: Date, format: "2020-04-02" "2020-04-03" "2020-04-04" "2020-04-05" ...

models

'data.frame': obs. of 7 variables:

p: num [0;1] proportion of confirmed cases
beta: num 13.8 13.8 16.3 11.5 29.2 ...
gamma: num 13.8 13.8 16.3 11.5 29.2 ...
CFR: num 0.14 0.14 0.2319 0.0312 0.0705 ...
cost: num 658 256 1207 1091 300 ...
region: chr, e.g., "Afghanistan/" "Albania/" "Algeria/" "Andorra/" ...

startSimulation

chr start of the simulation period, e.g., "2020-01-22". startSimulation must be at or before the Date from testData. Simulations can start earlier, because some use R=0. This enables a warm-up period.

write

logical. Default FALSE. If TRUE, results are written to the file submit.csv.

Value

returns data.frame with obs. of the following 3 variables:

ForecastId: int: Forecast Id taken from the regionTest data set.
ConfirmedCases: num: Cumulative number of confirmed cases.
Fatalities: num: Cumulative number of fatalities.

Details

Output from parseTunedRegionModel is processed-

Examples

Run this code

# NOT RUN {
require(SimInf)
data <- preprocessInputData(regionTrain, regionPopulation)
testData <- preprocessTestData(regionTest)
# Select the first region:
testData <- testData[testData$Region==levels(testData$Region)[1], ]
testData$Region <- droplevels(testData$Region)
# Very small number of function evaluations:
n <- 6
res <- lapply(data[1], tuneRegionModel, pops=NULL, 
control=list(funEvals=n, designControl=list(size=5), model = buildLM))
parsedList <- parseTunedRegionModel(res)
pred <- generateMCPrediction(testData = testData, models = parsedList$models, write = FALSE)
# }