GenerateR.java

import java.io.*;
import java.util.ArrayList;
import java.util.List;

/**
 * @author Chathura Widanage
 */
public class GenerateR {
    public static void main(String[] args) throws IOException {
        String pathToTrainCSV = "F:\\\\engineering\\\\Semester 7_8\\\\ML\\\\keggle\\\\train.csv";//4 backward slashes required
        File rScript = new File("D:\\cor.R");//file to save the generated R script
        PrintWriter pw = new PrintWriter(new FileWriter(rScript));
        pw.println("library(data.table)");
        pw.println("print('Reading data')");
        pw.printf("train <-fread('%s',select = c('Cliente_ID', 'Producto_ID', 'Agencia_ID', 'Ruta_SAK', 'Demanda_uni_equil','Canal_ID'))\n"
                , pathToTrainCSV);
        ArrayList<String> permeationStrings = new ArrayList<>();
        for (int i = 1; i < 32; i++) {//generating all possible combinations
            permeationStrings.add(String.format("%05d", Integer.parseInt(Integer.toBinaryString(i))));
        }
        String columns[] = {"Cliente_ID", "Producto_ID", "Agencia_ID", "Ruta_SAK", "Canal_ID"};
        permeationStrings.forEach(perm -> {
            char[] bits = perm.toCharArray();
            ArrayList<String> selectedColumns = new ArrayList<>();
            for (int i = 0; i < bits.length; i++) {
                if (bits[i] == '1') {
                    selectedColumns.add(columns[i]);
                }
            }
            String varName = getMeanVarName(selectedColumns);
            pw.printf("%s <-  train[, .(%s = mean(Demanda_uni_equil)), by = .(%s)]\n"
                    , varName, varName, getColString(selectedColumns, false));
            pw.printf("train <- merge(train, %s, all.x = TRUE, by = c(%s))\n"
                    , varName, getColString(selectedColumns, true));
            pw.printf("print(paste(cor(train$Demanda_uni_equil,train$%s),\"%s\"))\n", varName, " -> " + getColString(selectedColumns, false));
            pw.printf("train$%s <- NULL\n", varName);//cleaing
            pw.printf("%s <- NULL\n", varName);//cleaning
            pw.printf("gc()\n");//garbage collecting to prevent out of memory errors
        });
        pw.flush();
        pw.close();
    }

    public static String getColString(List<String> selectedCols, boolean withQuotes) {
        String cols = "";
        for (int i = 0; i < selectedCols.size(); i++) {
            String col = selectedCols.get(i);
            cols += withQuotes ? "\"" + col + "\"" : col;
            if (i != selectedCols.size() - 1) cols += ",";
        }
        return cols;
    }

    public static String getMeanVarName(List<String> selectedCols) {
        String colName = "mean_";
        for (String col : selectedCols) {
            colName += col.substring(0, 3);
        }
        return colName;
    }
}