Java Code Examples for tech.tablesaw.api.IntColumn#set()

public static void main(String[] args) throws Exception {

    Table property = Table.read().csv("../data/sacramento_real_estate_transactions.csv");

    IntColumn sqft = property.intColumn("sq__ft");
    IntColumn price = property.intColumn("price");

    sqft.set(sqft.isEqualTo(0), IntColumnType.missingValueIndicator());
    price.set(price.isEqualTo(0), IntColumnType.missingValueIndicator());

    Plot.show(Histogram.create("Distribution of prices", property.numberColumn("price")));

    Layout layout = Layout.builder().title("Distribution of property sizes").build();
    HistogramTrace trace =
        HistogramTrace.builder(property.numberColumn("sq__ft"))
            .marker(Marker.builder().color("#B10DC9").opacity(.70).build())
            .build();
    Plot.show(new Figure(layout, trace));

    Plot.show(Histogram2D.create("Distribution of price and size", property, "price", "sq__ft"));

    Plot.show(BoxPlot.create("Prices by property type", property, "type", "price"));
  }
 

private IntColumn assignToGroupsByStep(int step) {
  IntColumn groupColumn = IntColumn.create(GROUP_COL_TEMP_NAME, temp.rowCount());
  temp.addColumns(groupColumn);

  int groupId = 1;
  int withinGroupCount = 0;
  Row row = new Row(temp);

  while (row.hasNext()) {
    row.next();
    if (withinGroupCount < step) {
      withinGroupCount++;
      groupColumn.set(row.getRowNumber(), groupId);
    } else {
      groupId++;
      groupColumn.set(row.getRowNumber(), groupId);
      withinGroupCount = 1;
    }
  }
  int lastGroupSize = temp.where(numberColumn(GROUP_COL_TEMP_NAME).isEqualTo(groupId)).rowCount();
  if (lastGroupSize < step) {
    temp = temp.dropWhere(numberColumn(GROUP_COL_TEMP_NAME).isEqualTo(groupId));
  }
  temp.addColumns(IntColumn.indexColumn("index", temp.rowCount(), 1));
  return groupColumn;
}
 

public static void main(String[] args) throws Exception {

    Table property = Table.read().csv("../data/sacramento_real_estate_transactions.csv");

    IntColumn sqft = property.intColumn("sq__ft");
    IntColumn price = property.intColumn("price");

    sqft.set(sqft.isEqualTo(0), IntColumnType.missingValueIndicator());
    price.set(price.isEqualTo(0), IntColumnType.missingValueIndicator());

    Plot.show(Histogram.create("Distribution of prices", property.numberColumn("price")));

    Layout layout = Layout.builder().title("Distribution of property sizes").build();
    HistogramTrace trace =
        HistogramTrace.builder(property.numberColumn("sq__ft"))
            .marker(Marker.builder().color("#B10DC9").opacity(.70).build())
            .build();
    Plot.show(new Figure(layout, trace));

    Plot.show(Histogram2D.create("Distribution of price and size", property, "price", "sq__ft"));

    Plot.show(BoxPlot.create("Prices by property type", property, "type", "price"));
  }
 

default IntColumn minute() {
  IntColumn newColumn = IntColumn.create(name() + "[" + "minute" + "]", size());
  for (int r = 0; r < size(); r++) {
    if (!isMissing(r)) {
      long c1 = getLongInternal(r);
      newColumn.set(r, getMinute(c1));
    }
  }
  return newColumn;
}
 

default IntColumn dayOfYear() {
  IntColumn newColumn = IntColumn.create(this.name() + " day of year", this.size());
  for (int r = 0; r < this.size(); r++) {
    if (!isMissing(r)) {
      long c1 = this.getLongInternal(r);
      newColumn.set(r, (short) getDayOfYear(c1));
    }
  }
  return newColumn;
}
 

default IntColumn dayOfWeekValue() {
  IntColumn newColumn = IntColumn.create(this.name() + " day of week value", this.size());
  for (int r = 0; r < this.size(); r++) {
    if (!isMissing(r)) {
      long c1 = this.getLongInternal(r);
      newColumn.set(r, (short) getDayOfWeek(c1).getValue());
    }
  }
  return newColumn;
}
 

/**
 * Returns a column containing integers representing the nth group (0-based) that a date falls
 * into.
 *
 * <p>Example: When Unit = ChronoUnit.DAY and n = 5, we form 5 day groups. a Date that is 2 days
 * after the start is assigned to the first ("0") group. A day 7 days after the start is assigned
 * to the second ("1") group.
 *
 * @param unit A ChronoUnit greater than or equal to a day
 * @param n The number of units in each group.
 * @param start The starting point of the first group; group boundaries are offsets from this
 *     point
 */
default IntColumn timeWindow(ChronoUnit unit, int n, LocalDate start) {
  String newColumnName = "" + n + " " + unit.toString() + " window [" + name() + "]";
  int packedStartDate = PackedLocalDate.pack(start);
  IntColumn numberColumn = IntColumn.create(newColumnName, size());
  for (int i = 0; i < size(); i++) {
    int packedDate = getIntInternal(i);
    int result;
    switch (unit) {
      case DAYS:
        result = PackedLocalDate.daysUntil(packedDate, packedStartDate) / n;
        numberColumn.set(i, result);
        break;
      case WEEKS:
        result = PackedLocalDate.weeksUntil(packedDate, packedStartDate) / n;
        numberColumn.set(i, result);
        break;
      case MONTHS:
        result = PackedLocalDate.monthsUntil(packedDate, packedStartDate) / n;
        numberColumn.set(i, result);
        break;
      case YEARS:
        result = PackedLocalDate.yearsUntil(packedDate, packedStartDate) / n;
        numberColumn.set(i, result);
        break;
      default:
        throw new UnsupportedTemporalTypeException(
            "The ChronoUnit " + unit + " is not supported for timeWindows on dates");
    }
  }
  numberColumn.setPrintFormatter(NumberColumnFormatter.ints());
  return numberColumn;
}
 

default IntColumn dayOfWeekValue() {
  IntColumn newColumn = IntColumn.create(this.name() + " day of week", this.size());
  for (int r = 0; r < this.size(); r++) {
    int c1 = this.getIntInternal(r);
    if (DateColumn.valueIsMissing(c1)) {
      newColumn.setMissing(r);
    } else {
      newColumn.set(r, (short) PackedLocalDate.getDayOfWeek(c1).getValue());
    }
  }
  return newColumn;
}
 

default IntColumn dayOfMonth() {
  IntColumn newColumn = IntColumn.create(this.name() + " day of month", size());
  for (int r = 0; r < this.size(); r++) {
    if (!isMissing(r)) {
      long c1 = this.getLongInternal(r);
      newColumn.set(r, getDayOfMonth(c1));
    }
  }
  return newColumn;
}
 

public static void main(String[] args) throws Exception {

    Table tornadoes = Table.read().csv("../data/tornadoes_1950-2014.csv");
    out(tornadoes.structure());
    tornadoes.setName("tornadoes");

    // filter out a bad data point
    tornadoes = tornadoes.where(tornadoes.numberColumn("Start Lat").isGreaterThan(20f));

    IntColumn scale = tornadoes.intColumn("scale");
    scale.set(scale.isEqualTo(-9), IntColumnType.missingValueIndicator());

    Table fatalities1 = tornadoes.summarize("fatalities", sum).by("scale");

    Plot.show(
        HorizontalBarPlot.create(
            "Total fatalities by scale", fatalities1, "scale", "sum [fatalities]"));

    Plot.show(
        PiePlot.create("Total fatalities by scale", fatalities1, "scale", "sum [fatalities]"));

    Table fatalities2 = tornadoes.summarize("fatalities", sum).by("state");

    Plot.show(
        ParetoPlot.createVertical(
            "Total Tornado Fatalities by State", fatalities2, "state", "sum [fatalities]"));

    Table injuries1 = tornadoes.summarize("injuries", mean).by("scale");
    Plot.show(
        HorizontalBarPlot.create(
            "Tornado Injuries by Scale", injuries1, "scale", "mean [injuries]"));
    out(injuries1);
  }
 

default IntColumn dayOfMonth() {
  IntColumn newColumn = IntColumn.create(this.name() + " day of month", size());
  for (int r = 0; r < this.size(); r++) {
    if (!isMissing(r)) {
      long c1 = this.getLongInternal(r);
      newColumn.set(r, getDayOfMonth(c1));
    }
  }
  return newColumn;
}
 

default IntColumn dayOfYear() {
  IntColumn newColumn = IntColumn.create(this.name() + " day of year", this.size());
  for (int r = 0; r < this.size(); r++) {
    if (!isMissing(r)) {
      long c1 = this.getLongInternal(r);
      newColumn.set(r, (short) getDayOfYear(c1));
    }
  }
  return newColumn;
}
 

default IntColumn dayOfWeekValue() {
  IntColumn newColumn = IntColumn.create(this.name() + " day of week value", this.size());
  for (int r = 0; r < this.size(); r++) {
    if (!isMissing(r)) {
      long c1 = this.getLongInternal(r);
      newColumn.set(r, (short) getDayOfWeek(c1).getValue());
    }
  }
  return newColumn;
}
 

/**
 * Returns a column containing integers representing the nth group (0-based) that a date falls
 * into.
 *
 * <p>Example: When Unit = ChronoUnit.DAY and n = 5, we form 5 day groups. a Date that is 2 days
 * after the start is assigned to the first ("0") group. A day 7 days after the start is assigned
 * to the second ("1") group.
 *
 * @param unit A ChronoUnit greater than or equal to a day
 * @param n The number of units in each group.
 * @param start The starting point of the first group; group boundaries are offsets from this
 *     point
 */
default IntColumn timeWindow(ChronoUnit unit, int n, LocalDate start) {
  String newColumnName = "" + n + " " + unit.toString() + " window [" + name() + "]";
  int packedStartDate = PackedLocalDate.pack(start);
  IntColumn numberColumn = IntColumn.create(newColumnName, size());
  for (int i = 0; i < size(); i++) {
    int packedDate = getIntInternal(i);
    int result;
    switch (unit) {
      case DAYS:
        result = PackedLocalDate.daysUntil(packedDate, packedStartDate) / n;
        numberColumn.set(i, result);
        break;
      case WEEKS:
        result = PackedLocalDate.weeksUntil(packedDate, packedStartDate) / n;
        numberColumn.set(i, result);
        break;
      case MONTHS:
        result = PackedLocalDate.monthsUntil(packedDate, packedStartDate) / n;
        numberColumn.set(i, result);
        break;
      case YEARS:
        result = PackedLocalDate.yearsUntil(packedDate, packedStartDate) / n;
        numberColumn.set(i, result);
        break;
      default:
        throw new UnsupportedTemporalTypeException(
            "The ChronoUnit " + unit + " is not supported for timeWindows on dates");
    }
  }
  numberColumn.setPrintFormatter(NumberColumnFormatter.ints());
  return numberColumn;
}
 

default IntColumn dayOfWeekValue() {
  IntColumn newColumn = IntColumn.create(this.name() + " day of week", this.size());
  for (int r = 0; r < this.size(); r++) {
    int c1 = this.getIntInternal(r);
    if (DateColumn.valueIsMissing(c1)) {
      newColumn.setMissing(r);
    } else {
      newColumn.set(r, (short) PackedLocalDate.getDayOfWeek(c1).getValue());
    }
  }
  return newColumn;
}
 

public static void main(String[] args) throws Exception {

    Table tornadoes = Table.read().csv("../data/tornadoes_1950-2014.csv");
    out(tornadoes.structure());
    tornadoes.setName("tornadoes");

    // filter out a bad data point
    tornadoes = tornadoes.where(tornadoes.numberColumn("Start Lat").isGreaterThan(20f));

    IntColumn scale = tornadoes.intColumn("scale");
    scale.set(scale.isEqualTo(-9), IntColumnType.missingValueIndicator());

    Table fatalities1 = tornadoes.summarize("fatalities", sum).by("scale");

    Plot.show(
        HorizontalBarPlot.create(
            "Total fatalities by scale", fatalities1, "scale", "sum [fatalities]"));

    Plot.show(
        PiePlot.create("Total fatalities by scale", fatalities1, "scale", "sum [fatalities]"));

    Table fatalities2 = tornadoes.summarize("fatalities", sum).by("state");

    Plot.show(
        ParetoPlot.createVertical(
            "Total Tornado Fatalities by State", fatalities2, "state", "sum [fatalities]"));

    Table injuries1 = tornadoes.summarize("injuries", mean).by("scale");
    Plot.show(
        HorizontalBarPlot.create(
            "Tornado Injuries by Scale", injuries1, "scale", "mean [injuries]"));
    out(injuries1);
  }
 

default IntColumn minute() {
  IntColumn newColumn = IntColumn.create(name() + "[" + "minute" + "]", size());
  for (int r = 0; r < size(); r++) {
    if (!isMissing(r)) {
      long c1 = getLongInternal(r);
      newColumn.set(r, getMinute(c1));
    }
  }
  return newColumn;
}
 

public static void main(String[] args) throws Exception {
  Table table = Table.read().csv("../data/tornadoes_1950-2014.csv");
  NumericColumn<?> logNInjuries = table.numberColumn("injuries").add(1).logN();
  logNInjuries.setName("log injuries");
  table.addColumns(logNInjuries);
  IntColumn scale = table.intColumn("scale");
  scale.set(scale.isLessThan(0), IntColumnType.missingValueIndicator());

  Table summaryTable = table.summarize("fatalities", "log injuries", sum).by("Scale");

  Plot.show(
      HorizontalBarPlot.create(
          "Tornado Impact",
          summaryTable,
          "scale",
          Layout.BarMode.STACK,
          "Sum [Fatalities]",
          "Sum [log injuries]"));

  Plot.show(
      VerticalBarPlot.create(
          "Tornado Impact",
          summaryTable,
          "scale",
          Layout.BarMode.GROUP,
          "Sum [Fatalities]",
          "Sum [log injuries]"));

  Layout layout =
      Layout.builder()
          .title("Tornado Impact")
          .barMode(Layout.BarMode.GROUP)
          .showLegend(true)
          .build();

  String[] numberColNames = {"Sum [Fatalities]", "Sum [log injuries]"};
  String[] colors = {"#85144b", "#FF4136"};

  Trace[] traces = new Trace[2];
  for (int i = 0; i < 2; i++) {
    String name = numberColNames[i];
    BarTrace trace =
        BarTrace.builder(summaryTable.categoricalColumn("scale"), summaryTable.numberColumn(name))
            .orientation(BarTrace.Orientation.VERTICAL)
            .marker(Marker.builder().color(colors[i]).build())
            .showLegend(true)
            .name(name)
            .build();
    traces[i] = trace;
  }
  Plot.show(new Figure(layout, traces));
}
 

public static void main(String[] args) throws Exception {
  Table table = Table.read().csv("../data/tornadoes_1950-2014.csv");
  NumericColumn<?> logNInjuries = table.numberColumn("injuries").add(1).logN();
  logNInjuries.setName("log injuries");
  table.addColumns(logNInjuries);
  IntColumn scale = table.intColumn("scale");
  scale.set(scale.isLessThan(0), IntColumnType.missingValueIndicator());

  Table summaryTable = table.summarize("fatalities", "log injuries", sum).by("Scale");

  Plot.show(
      HorizontalBarPlot.create(
          "Tornado Impact",
          summaryTable,
          "scale",
          Layout.BarMode.STACK,
          "Sum [Fatalities]",
          "Sum [log injuries]"));

  Plot.show(
      VerticalBarPlot.create(
          "Tornado Impact",
          summaryTable,
          "scale",
          Layout.BarMode.GROUP,
          "Sum [Fatalities]",
          "Sum [log injuries]"));

  Layout layout =
      Layout.builder()
          .title("Tornado Impact")
          .barMode(Layout.BarMode.GROUP)
          .showLegend(true)
          .build();

  String[] numberColNames = {"Sum [Fatalities]", "Sum [log injuries]"};
  String[] colors = {"#85144b", "#FF4136"};

  Trace[] traces = new Trace[2];
  for (int i = 0; i < 2; i++) {
    String name = numberColNames[i];
    BarTrace trace =
        BarTrace.builder(summaryTable.categoricalColumn("scale"), summaryTable.numberColumn(name))
            .orientation(BarTrace.Orientation.VERTICAL)
            .marker(Marker.builder().color(colors[i]).build())
            .showLegend(true)
            .name(name)
            .build();
    traces[i] = trace;
  }
  Plot.show(new Figure(layout, traces));
}
 

public static void main(String[] args) throws Exception {

    // ***************** Setup *************************
    // load the data into a table
    Table tornadoes = Table.read().csv("../data/tornadoes_1950-2014.csv");

    // Get the scale column and replace any values of -9 with the column's missing value indicator
    IntColumn scale = tornadoes.intColumn("scale");
    scale.set(scale.isEqualTo(-9), IntColumnType.missingValueIndicator());

    // ***************** Plotting **********************

    // BAR PLOTS

    // Sum the number of fatalities from each tornado, grouping by scale
    Table fatalities1 = tornadoes.summarize("fatalities", sum).by("scale");

    // Plot
    Plot.show(
        HorizontalBarPlot.create(
            "fatalities by scale", // plot title
            fatalities1, // table
            "scale", // grouping column name
            "sum [fatalities]")); // numeric column name

    // Plot the mean injuries rather than a sum.
    Table injuries1 = tornadoes.summarize("injuries", mean).by("scale");

    Plot.show(
        HorizontalBarPlot.create(
            "Average number of tornado injuries by scale", injuries1, "scale", "mean [injuries]"));

    // PIE PLOT
    Plot.show(PiePlot.create("fatalities by scale", fatalities1, "scale", "sum [fatalities]"));

    // PARETO PLOT
    Table t2 = tornadoes.summarize("fatalities", sum).by("State");

    t2 = t2.sortDescendingOn(t2.column(1).name());
    Layout layout = Layout.builder().title("Tornado Fatalities by State").build();
    BarTrace trace = BarTrace.builder(t2.categoricalColumn(0), t2.numberColumn(1)).build();
    Plot.show(new Figure(layout, trace));
  }