prediction_accuracy.Rmd

---
title: "WWW2017"
author: Pavel Kucherbaev, Florian Daniel, Maurizio Marchese
date: September 22, 2016
output:
  md_document:
    variant: markdown_github
---

```{r}
library(sqldf)
library(caret)
library(rpart)
library(rpart.plot)
require(dplyr)
```

```{r, echo = TRUE, width = 20, height = 20}
source("Approach/tasks_for_simulation.r")
source("Libs/logs_into_features.r")
source("Approach/simulations_lib.r")
#source("Approach/simulations_lib_linear.r")
task_list <- getSimulationTaskList()

task1 <- prepareFeaturesAssignmentsDataset(task_list[1,1], task_list[1,3], task_list[1,2])
task1$re_evaluation <- as.factor(task1$re_evaluation)

task2 <- prepareFeaturesAssignmentsDataset(task_list[2,1], task_list[2,3], task_list[2,2])
task2$re_evaluation <- as.factor(task2$re_evaluation)

task_2_features <- prepareFeaturesAssignmentsDataset(task_list[2,1],task_list[2,3], task_list[2,2])
task_2_features <- prepareEvaluationTrainingSet(task_2_features)
task_4_features <- prepareFeaturesAssignmentsDataset(task_list[4,1],task_list[4,3], task_list[4,2])
task_4_features <- prepareEvaluationTrainingSet(task_4_features)
task_key_features <- prepareKeysActivityLogs(task_list[2,1])

# DO NOT use tasks 1 and 4
training_task = 8
test_task = 9
simulation <- simulate_speed_ml(task_list[test_task,1], task_list[test_task,2],task_list[training_task,1], task_list[training_task,2],1)

for(i in seq(1:9)){
	simulate_speed_linear(task_list[i,1], task_list[i,2])
}

for(task_type in c(1,4,7)){
	for(training_task in c(0,1,2)){
	  for(test_task in c(0,1,2)){
	    print(paste(task_type,training_task,test_task,sep=" "))
	    if (test_task != training_task){
  		  simulate_accuracy_ml(task_list[task_type+test_task,1], task_list[task_type+test_task,2],task_list[task_type+training_task,1], task_list[task_type+training_task,2],0.2)
	    }
	  }
	}
}
```

```{r}
SIMULATION_TASKS
```