package quickml.Utilities;
import au.com.bytecode.opencsv.CSVReader;
import com.google.common.collect.Lists;
import com.google.common.collect.Maps;
import quickml.supervised.regressionModel.IsotonicRegression.PoolAdjacentViolatorsModel;
import java.io.FileReader;
import java.util.*;
/**
* Created by alexanderhawk on 10/2/14.
*/
/* This class converts the contents of a csv file into a map of lists...where the contents of each collumn are
stored in a List. Each list is contained in a map, where the column names (or numbers if names are not
provided in a header) are the keys for their respective lists.
*/
public class CSVToMapOfNumericLists {
private List<String> header;
private char delimiter = ',';
private boolean headerPresent = true;
private Map<String, List<Double>> dataLists = Maps.newHashMap();
public CSVToMapOfNumericLists() {
}
public CSVToMapOfNumericLists(char delimiter) {
this.delimiter = delimiter;
}
public CSVToMapOfNumericLists(char delimiter, boolean noHeaderPresent) {
this.delimiter = delimiter;
this.headerPresent = noHeaderPresent;
}
public Map<String, List<Double>> readCsv(String fileName) throws Exception {
CSVReader reader = new CSVReader(new FileReader(fileName), delimiter, '"');
List<String[]> csvLines = reader.readAll();
try {
header = Lists.newArrayList();
int startIndex = 1;
if (!headerPresent) {
startIndex = 0;
for (int i = 0; i < csvLines.get(0).length; i++) {
header.add(String.valueOf(i));
dataLists.put(header.get(i), Lists.<Double>newArrayList());
}
} else {
for (int i = 0; i < csvLines.get(0).length; i++) {
header.add(csvLines.get(0)[i]);
dataLists.put(header.get(i), Lists.<Double>newArrayList());
}
for (int i = startIndex; i < csvLines.size(); i++) {
appendLineToLists(csvLines.get(i));
}
}
} catch (Exception e) {
throw new RuntimeException(e.getMessage());
}
return dataLists;
}
private void appendLineToLists(String[] dataLine) {
for (int i = 0; i < dataLine.length; i++) {
if (dataLine[i].isEmpty()) {
continue;
}
List<Double> listToappendTo = dataLists.get(header.get(i));
try {
listToappendTo.add(Double.valueOf(dataLine[i]));
} catch (NumberFormatException e) {
listToappendTo.add(Double.NaN);
}
}
}
public static void main(String[] args) {
CSVToMapOfNumericLists csvReader = new CSVToMapOfNumericLists(',', true);
try {
Map<String, List<Double>> mapOfinstances = csvReader.readCsv("wfRes");
for (String key : mapOfinstances.keySet()) {
System.out.println("list key: " + key + "list Vals" + mapOfinstances.get(key).toString());
}
} catch (Exception e) {
throw new RuntimeException();
}
List<PoolAdjacentViolatorsModel.Observation> observations = Lists.newArrayList();
List<PoolAdjacentViolatorsModel.Observation> predictions = Lists.newArrayList();
for (int i = 0; i < csvReader.dataLists.get("target_cpc_customers").size(); i++) {
PoolAdjacentViolatorsModel.Observation observation = new PoolAdjacentViolatorsModel.Observation
(csvReader.dataLists.get("target_cpc_customers").get(i), csvReader.dataLists.get("effective_cust_daily_spend").get(i));
observations.add(observation);
}
PoolAdjacentViolatorsModel pav = new PoolAdjacentViolatorsModel(observations, 4);
PoolAdjacentViolatorsModel.Observation prev = null;
// observations.sort(Comparator.<PoolAdjacentViolatorsModel.Observation>naturalOrder());
for (PoolAdjacentViolatorsModel.Observation observation : observations) {
PoolAdjacentViolatorsModel.Observation obsToAdd = new PoolAdjacentViolatorsModel.Observation(observation.input, pav.predict(observation.input));
predictions.add(obsToAdd);
if (prev !=null && prev.output < observation.output)
System.out.println("prev out: " + prev.toString() + ". obs.input: " + obsToAdd.toString());
prev = obsToAdd;
}
TreeSet<PoolAdjacentViolatorsModel.Observation> calibrationSet = pav.getCalibrationSet();
TreeSet<PoolAdjacentViolatorsModel.Observation> preSmoothingSet = pav.getPreSmoothingSet();
System.out.println("calibration set: " + calibrationSet.toString());
LinePlotter linePlotter = new LinePlotterBuilder().chartTitle("pav trial").xAxisLabel("cpc").yAxisLabel("rate").buildLinePlotter();
linePlotter.addSeries(calibrationSet, "unweigted linear interpolation");
linePlotter.addSeries(predictions, "weighted linear interpolation");
linePlotter.addSeries(preSmoothingSet, "binned observations with weight 4");
linePlotter.displayPlot();
}
}