C＃/ Excel：解决图表上的最大序列大小

Question

I need help programatically graphing more points than can fit in a single Excel series.

According to http://office.microsoft.com/en-us/excel/HP100738491033.aspx the maximum number of points displayable on an Excel 2007 chart is 256000. Given that each series caps out at 32000 points, 8 series are required to plot the full 256000 points. My customer requires plotting of maximum amount of points per chart due to the large data sets we work with.

I have moderate experience with C#/Excel interop so I thought it would be easy to programatically create a worksheet and then loop through each set of 32000 points and add them to the graph as a series, stopping when the data was fully plotted or 8 series were plotted. If colored properly, the 8 series would be visually indistinguishable from a single series.

Unfortunately here I am. The main problem I encounter is:

(full size) The maximum number of datapoints you can use in a data series for a 2-D chart is 32,000... http://img14.imageshack.us/img14/9630/errormessagen.png

This pop-up, strangely enough, appears when I execute the line:

and is accompanied by:

Exception from HRESULT: 0x800AC472 http://img21.imageshack.us/img21/5153/exceptionb.png

I do not understand how I could be generating such a popup/warning/exception before I have even specified the data to be graphed. Is Excel trying to be clever here?

As a temporary workaround, I've put the chart.ChartType = chartType statement into a try-catch block so I can keep going.

As the following shows, my "chunking" code is working as intended, but I still encounter the same problem when trying to add data to the graph. Excel says I am trying to graph too many points when clearly I am not.

( full size image ) code block with watch window http://img12.imageshack.us/img12/5360/snippet.png

I understand I may not have the X Values correctly associated with each series yet, but I'm trying to get this to work before I go further.

Any help would be greatly appreciated.

Here's the full code:

public void DrawScatterGraph(string xColumnLetter, string yColumnLetterStart, string yColumnLetterStop, string xAxisLabel, string yAxisLabel, string chartTitle, Microsoft.Office.Interop.Excel.XlChartType chartType, bool includeTrendline, bool includeLegend)
    {
        int totalRows = dataSheet.UsedRange.Rows.Count; //dataSheet is a private class variable that 
                                                        //is already properly set to the worksheet
                                                        //we want to graph from

        if (totalRows < 2) throw new Exception("Not generating graph for " + chartTitle.Replace('\n', ' ') 
                                            + " because not enough data was present");

        ChartObjects charts = (ChartObjects)dataSheet.ChartObjects(Type.Missing);
        ChartObject chartObj = charts.Add(100, 300, 500, 300);
        Chart chart = chartObj.Chart;

        try { chart.ChartType = chartType; }
        catch { }   //i don't know why this is throwing an exception, but i'm
                    //going to bulldoze through this problem temporarily 

        if (totalRows < SizeOfSeries) //we can graph the data in a single series - yay!
        {
            Range xValues = dataSheet.get_Range(xColumnLetter + "2", xColumnLetter + totalRows.ToString());
            Range yValues = dataSheet.get_Range(yColumnLetterStart + "1", yColumnLetterStop + totalRows.ToString());
            chart.SetSourceData(yValues, XlRowCol.xlColumns);
            SeriesCollection seriesCollection = (SeriesCollection)chart.SeriesCollection(Type.Missing);
            foreach (Series s in seriesCollection)
            {
                s.XValues = xValues;
            }
        }
        else // we need to split the data across multiple series -- this doesn't work yet
        {
            int startRow = 1; 
            while (startRow < totalRows)
            {
                int stopRow = (startRow + SizeOfSeries)-1;  
                if (stopRow > totalRows) stopRow = totalRows;
                Range curRange = dataSheet.get_Range(yColumnLetterStart + startRow.ToString(), yColumnLetterStop + stopRow.ToString());
                try
                {
                    ((SeriesCollection)chart.SeriesCollection(Type.Missing)).Add(curRange, XlRowCol.xlColumns, 
                                                                            Type.Missing, Type.Missing, Type.Missing);
                }
                catch (Exception exc)
                {
                    throw new Exception(yColumnLetterStart + startRow.ToString() + "!" + yColumnLetterStop + stopRow.ToString() + "!" + exc.Message);
                }
                startRow = stopRow+1;
            }
        }

        chart.HasLegend = includeLegend;
        chart.HasTitle = true;
        chart.ChartTitle.Text = chartTitle;

        Axis axis;
        axis = (Axis)chart.Axes(XlAxisType.xlCategory, XlAxisGroup.xlPrimary);
        axis.HasTitle = true;
        axis.AxisTitle.Text = xAxisLabel;
        axis.HasMajorGridlines = false;
        axis.HasMinorGridlines = false;

        axis = (Axis)chart.Axes(XlAxisType.xlValue, XlAxisGroup.xlPrimary);
        axis.HasTitle = true;
        axis.AxisTitle.Text = yAxisLabel;
        axis.HasMajorGridlines = true;
        axis.HasMinorGridlines = false;

        if (includeTrendline)
        {
            Trendlines t = (Trendlines)((Series)chart.SeriesCollection(1)).Trendlines(Type.Missing);
            t.Add(XlTrendlineType.xlLinear, Type.Missing, Type.Missing, 0, 0, Type.Missing, false, false, "AutoTrendlineByChameleon");
        }

        chart.Location(XlChartLocation.xlLocationAsNewSheet, "Graph");
    }

Answer 1

Does your graph actually have to be in Excel? With that many data points the performance would be horrible.

One suggestion might be to use a third party component to generate the graph. The specific technique for how to accomplish this depends on whether you have to be able to view the data in excel or whether the output graph simply needs to be available elsewhere.

If the graph does not need to be visible within Excel, then just pass the data points and view the image in the graphing application or a web browser.

If you do need to view the graph with excel, you could make a call to the external graphing application and pass it a collection of data points. When it returns the image just insert it in excel with vba.

I can give you more info on both approaches if you need.

Also, other considerations might include whether you need to have drill down capability on the graph. With this many data points, I can not imagine that you would.

---

If you can answer the following questions, it might help folks formulate better answers.

1. What sort of user interface will be presenting the output of these items? (eg Excel, ASP.NET Web Application, Windows Forms, WPF, Silverlight, other.)

2. Are these graphs supposed to be generated in real time at a user's request or are they generated and stored? If they are generated on demand, what is the maximum amount of time your users would consider acceptable to wait?

3. How important is it that you actually use Excel? Are you using it because it is a requirement for display, or is that just what is handy?

4. How important is the "Wow factor" for the display of the graphs? Is simply having the graphs, or do they have to be extremely beautiful?

5. Do users require any ability to drill down into the graph, or is simply being able to view the image sufficient?

Answer 2

If the active cell is in a block of data, Excel may assume you want to plot the range.

Select a blank cell which is not next to the data, then insert the chart. It will be blank, rather than prepopulated.

Answer 3

To help anyone who comes across this in the future, here's the complete function with Jon's fix:

    public void DrawScatterGraph(string xColumnLetter, string yColumnLetterStart, string yColumnLetterStop, string xAxisLabel, string yAxisLabel, string chartTitle, Microsoft.Office.Interop.Excel.XlChartType chartType, bool includeTrendline, bool includeLegend)
    {
        int totalRows = dataSheet.UsedRange.Rows.Count; //dataSheet is a private class variable that 
                                                        //is already properly set to the worksheet
                                                        //we want to graph from

        if (totalRows < 2) throw new Exception("Not generating graph for " + chartTitle.Replace('\n', ' ') 
                                               + " because not enough data was present");

        dataSheet.get_Range("Z1", "Z2").Select();   //we need to select some empty space
                                                    //so Excel doesn't try to jam the 
                                                    //potentially large data set into the 
                                                    //chart automatically

        ChartObjects charts = (ChartObjects)dataSheet.ChartObjects(Type.Missing);
        ChartObject chartObj = charts.Add(100, 300, 500, 300);
        Chart chart = chartObj.Chart;
        chart.ChartType = chartType;
        SeriesCollection seriesCollection = (SeriesCollection)chart.SeriesCollection(Type.Missing);

        if (totalRows < SizeOfSeries) //we can graph the data in a single series - yay!
        {
            Range xValues = dataSheet.get_Range(xColumnLetter + "2", xColumnLetter + totalRows.ToString());
            Range yValues = dataSheet.get_Range(yColumnLetterStart + "1", yColumnLetterStop + totalRows.ToString());
            chart.SetSourceData(yValues, XlRowCol.xlColumns);

            foreach (Series s in seriesCollection)
            {
                s.XValues = xValues;
            }
        }
        else // we need to split the data across multiple series 
        {
            int startRow = 2; 

            while (startRow < totalRows)
            {
                int stopRow = (startRow + SizeOfSeries)-1;  
                if (stopRow > totalRows) stopRow = totalRows;

                Series s = seriesCollection.NewSeries();
                s.Name = "ChunkStartingAt" + startRow.ToString();
                s.XValues = dataSheet.get_Range(xColumnLetter + startRow.ToString(), xColumnLetter + stopRow.ToString());
                s.Values = dataSheet.get_Range(yColumnLetterStart + startRow.ToString(), yColumnLetterStop + stopRow.ToString());

                startRow = stopRow+1;
            }
        }

        chart.HasLegend = includeLegend;
        chart.HasTitle = true;
        chart.ChartTitle.Text = chartTitle;

        Axis axis;
        axis = (Axis)chart.Axes(XlAxisType.xlCategory, XlAxisGroup.xlPrimary);
        axis.HasTitle = true;
        axis.AxisTitle.Text = xAxisLabel;
        axis.HasMajorGridlines = false;
        axis.HasMinorGridlines = false;

        axis = (Axis)chart.Axes(XlAxisType.xlValue, XlAxisGroup.xlPrimary);
        axis.HasTitle = true;
        axis.AxisTitle.Text = yAxisLabel;
        axis.HasMajorGridlines = true;
        axis.HasMinorGridlines = false;

        if (includeTrendline)
        {
            Trendlines t = (Trendlines)((Series)chart.SeriesCollection(1)).Trendlines(Type.Missing);
            t.Add(XlTrendlineType.xlLinear, Type.Missing, Type.Missing, 0, 0, Type.Missing, false, false, "AutoTrendlineByChameleon");
        }

        chart.Location(XlChartLocation.xlLocationAsNewSheet, "Graph");
    }