How can I change the value of 'StartRow' and 'EndRow' inside the loop?
for k = 1:11
filename = 'file.txt';
...
startRow = 1422; %1564, 1706, 1848, 1990, 2132, 2274, 2416, 2558, 2700, 2842
endRow = 1562; %1704, 1846, 1988, 2130, 2272, 2414, 2556, 2698, 2840, 2982
...
f=figure;
plot(...);
saveas(f,sprintf('fig%d.png',k));
end
You can store all values of startRow and endRow in a list right before the for loop and then iterate through the list to change the value of these variables inside the loop.
startRowList = [1422, 1564, 1706, 1848, 1990, 2132, 2274, 2416, 2558, 2700, 2842];
endRowList = [1562, 1704, 1846, 1988, 2130, 2272, 2414, 2556, 2698, 2840, 2982];
for k = 1:11
filename = 'file.txt';
...
startRow = startRowList(k);
endRow = endRowList(k);
...
f=figure;
plot(...);
saveas(f,sprintf('fig%d.png',k));
end
Keep startRow and endRow outside the loop and call them inside the loop using loop index.
startRow = [1422 1564, 1706, 1848, 1990, 2132, 2274, 2416, 2558, 2700, 2842] ;
endRow = [1562 1704, 1846, 1988, 2130, 2272, 2414, 2556, 2698, 2840, 2982];
n = length(startRow) ;
for k = 1:n
filename = 'file.txt';
thestart = startRow(k) ;
theend = endRow(k) ;
...
...
f=figure;
plot(...);
saveas(f,sprintf('fig%d.png',k));
end
starts = [1422, 1564, 1706, 1848, 1990, 2132, 2274, 2416, 2558, 2700, 2842];
ends = [1562, 1704, 1846, 1988, 2130, 2272, 2414, 2556, 2698, 2840, 2982];
for k = 1:11
...
startRow = starts(k);
endRow = ends(k);
...
end
Related
Somehow I cannot get my chart to ‘stack’. For reasons explained below I first prepare a standard chart in a macro called ‘ChartsGeneral’. Based on some settings on my ‘Charts’ worksheet, it then calls a specific second macro, e.g. ‘ChartBuildStatus’. In this second macro I generate the category and data-series, and adjust the settings of the standard chart as necessary.
When I try to get a stacked chart via ‘createInstance(“com.sun.star.chart.StackableDiagram”)’ nothing changes. I have studied the documentation and several forums, but I have no clue what I am missing. Code below.
Configuration details
Version: 7.4.3.2 (x64) / LibreOffice Community
CPU threads: 12; OS: Windows 10.0 Build 19045; Locale: nl-NL (nl_NL); UI: en-US
OK, here it goes… Please have some mercy on a poor old non-programmer nerd.
Background
Over the years I collected a rather large set of Excel spreadsheets with VBA macros for my personal finances. This year I decided to migrate them all to LibreOffice, as part of my wider migration from Windows to Linux. I am not a programmer, so I decided to use BASIC as it would be closest to VBA.
One of my spreadsheets tracks progress on our home mortgage. I keep track of all payments on interest and principal. As we are paying down extra with a fixed monthly payment and a ‘snowball’ (all savings go back as payments to the principal) the spreadsheet also calculates the expected next snowball payment, as well as a prognosis for the remainder of the mortgage. That all works fine.
The key worksheets in my Mortgage spreadsheet are ‘Months’, where I store all the data of past payments, and ‘Prognosis’, which holds a copy of ‘Months’ but also a prognosis for every month in the future.
I am now down to the last block of functionality: reporting. This is mainly a collection of charts that show the history, current status and prognosis for my mortgage based on the data in Months and Prognosis. For this I use a dedicated worksheet called ‘Charts’ in my spreadsheet. Here I can select a Chart Type from a dropdown-list, as well as some date or period related settings (I use named fields for this). See a partial screenshot below.
Partial Screenshot of Charts worksheet
When I press the ‘Update Chart’ button on my Charts worksheet, a macro called ‘ChartsGeneral’ is triggered. In this macro:
I use the date or period related settings to calculate the starting row and ending row for the data series
I then generate an empty standard linechart called ‘AdminChart’ on the Charts worksheet, with settings for the axes, the formatting, and everything
I call a second macro based on the ChartType value on Charts. In the above example the chart is called ‘Opbouw Stand van Zaken’, so I call a second macro called ChartBuildStatus.
The code for ChartsGeneral can be found below
Sub ChartsGeneral()
'Manages the Charts worksheet
'Set Worksheets
wshC = ThisComponent.Sheets.getByName("Charts")
wshM = ThisComponent.Sheets.getByName("Months")
wshP = ThisComponent.Sheets.getByName("Prognosis")
wshS = ThisComponent.Sheets.getByName("Settings")
'Define Colours
cBlack = RGB(0, 0, 0)
cGray01 = RGB(50, 50, 50)
cGray02 = RGB(191, 191, 191)
cWhite = RGB(255, 255, 255)
cDarkRed = RGB(192, 0, 0)
cPinkiDEAL = RGB(204, 0, 102) 'iDeal
cGreenCash = RGB(112, 173, 71) 'Cash Green
cBluePin = RGB(41, 65, 113) 'PIN Blue
cBlueContactless = RGB(58, 125, 193) 'Contactloos Light Blue
cOrangeVisa = RGB(246, 155, 14) 'Visa Orange
'Set Parameters
'Prognosis
ColumnOriginal = 18 'Original Rest Mortgage
ColumnRest = 17 'Actual Rest Mortgae
ColumnPaid = 16 'Reeds afgelost
ColumnPrognosis = 21 'Prognosis Rest Mortgage
ColumnInterestSavedPrognosis = 23 'Bespaarde rente prognose
ColumnInterestSaved = 26 'Bespaarde rente
ColumnExtraTotal = 14 'Column total extra cumulative
ColumnPercent = 24
ColumnPercentPrognosis = 25
'Months
XASColumn = 2
ColumnExtra = 13 'Extra repayment column
ColumnRegular = 12 'Regular repayment column
ColumnInterest = 11 'Interest payment column
ColumnMonthInterest = 3 'Interest payment column
ColumnMonthRegular = 4 'Regular repayment column
ColumnMonthTotal = 9 'Total Monthly repayment column
ColumnSnowball = 7 'Snowball per month column
ColumnWOZ = 17 'WOZ column
ColumnMortgage = 16 'Rest of mortgage column
ColumnVasteLasten = 5 'Regular Interest and Depreciation
ColumnLTV = 21 'LTV Ratio
ColumnPercentageRest = 24 'Percentage Rest Mortgage Paid This Month
ColumnWOZpercentage = 22 'WOZ percentage column
ColumnMortgagePercentage = 26 'Rest of mortgage percentage column
'X-As Values
LastMonthRow = getLastContentIndex("Months", 3)
If wshM.getCellByPosition(1, LastMonthRow).Value = (ThisComponent.NamedRanges.getByName("ChartMonth").getReferredCells.Value + 1) Then
LastMonthRow = LastMonthRow - 1
End If
LastMonth = wshM.getCellByPosition(1, LastMonthRow).Value
LastYear = wshM.getCellByPosition(0, LastMonthRow).Value
xStartRow = 3
xEndRow = LastMonthRow
If ThisComponent.NamedRanges.getByName("ChartPeriod").getReferredCells.String = "1Y" Then
StartYear = LastYear - 1
StartMonth = LastMonth + 1
If StartMonth = 13 Then
StartYear = StartYear + 1
StartMonth = 1
End If
For i = xStartRow To LastMonthRow
If wshM.getCellByPosition(0, i).Value = StartYear And wshM.getCellByPosition(1, i).Value = StartMonth Then
xStartRow = i
Exit For
End If
Next i
End If
If ThisComponent.NamedRanges.getByName("ChartPeriod").getReferredCells.String = "2Y" Then
StartYear = LastYear - 2
StartMonth = LastMonth + 1
If StartMonth = 13 Then
StartYear = StartYear + 1
StartMonth = 1
End If
For i = xStartRow To LastMonthRow
If wshM.getCellByPosition(0, i).Value = StartYear And wshM.getCellByPosition(1, i).Value = StartMonth Then
xStartRow = i
Exit For
End If
Next i
End If
If ThisComponent.NamedRanges.getByName("ChartPeriod").getReferredCells.String = "3Y" Then
StartYear = LastYear - 3
StartMonth = LastMonth + 1
If StartMonth = 13 Then
StartYear = StartYear + 1
StartMonth = 1
End If
For i = xStartRow To LastMonthRow
If wshM.getCellByPosition(0, i).Value = StartYear And wshM.getCellByPosition(1, i).Value = StartMonth Then
xStartRow = i
Exit For
End If
Next i
End If
If ThisComponent.NamedRanges.getByName("ChartPeriod").getReferredCells.String = "5Y" Then
StartYear = LastYear - 5
StartMonth = LastMonth + 1
If StartMonth = 13 Then
StartYear = StartYear + 1
StartMonth = 1
End If
For i = xStartRow To LastMonthRow
If wshM.getCellByPosition(0, i).Value = StartYear And wshM.getCellByPosition(1, i).Value = StartMonth Then
xStartRow = i
Exit For
End If
Next i
End If
If ThisComponent.NamedRanges.getByName("ChartPeriod").getReferredCells.String = "10Y" Then
StartYear = LastYear - 10
StartMonth = LastMonth + 1
If StartMonth = 13 Then
StartYear = StartYear + 1
StartMonth = 1
End If
For i = xStartRow To LastMonthRow
If wshM.getCellByPosition(0, i).Value = StartYear And wshM.getCellByPosition(1, i).Value = StartMonth Then
xStartRow = i
Exit For
End If
Next i
End If
If ThisComponent.NamedRanges.getByName("ChartPeriod").getReferredCells.String = "15Y" Then
StartYear = LastYear - 15
StartMonth = LastMonth + 1
If StartMonth = 13 Then
StartYear = StartYear + 1
StartMonth = 1
End If
For i = xStartRow To LastMonthRow
If wshM.getCellByPosition(0, i).Value = StartYear And wshM.getCellByPosition(1, i).Value = StartMonth Then
xStartRow = i
Exit For
End If
Next i
End If
If ThisComponent.NamedRanges.getByName("ChartPeriod").getReferredCells.String = "20Y" Then
StartYear = LastYear - 20
StartMonth = LastMonth + 1
If StartMonth = 13 Then
StartYear = StartYear + 1
StartMonth = 1
End If
For i = xStartRow To LastMonthRow
If wshM.getCellByPosition(0, i).Value = StartYear And wshM.getCellByPosition(1, i).Value = StartMonth Then
xStartRow = i
Exit For
End If
Next i
End If
'Global Variable for BlogMarkers
ChartStartRow = StartRow
'Generate New Chart
oCharts = wshC.Charts
If oCharts.Count <> 0 then
For i = 0 to oCharts.Count-1
oChart = oCharts.getByIndex(i)
If oChart.name = "AdminChart" then
oCharts.removeByName("AdminChart")
End If
Next i
End If
Dim oRange as Object
Dim oRangeAddress(1) As New com.sun.star.table.CellRangeAddress
Dim oRect As New com.sun.star.awt.Rectangle
Dim cTitle as String
'Define Point and Size in order to change Position and Size of Chart Objects
Dim Pos_Chart As New com.sun.star.awt.Point
Dim Pos_Title As New com.sun.star.awt.Point
Dim Pos_SubTitle As New com.sun.star.awt.Point
Dim Pos_xTitle As New com.sun.star.awt.Point
Dim Pos_yTitle As New com.sun.star.awt.Point
Dim Pos_Legend As New com.sun.star.awt.Point
Dim Size_Chart As New com.sun.star.awt.Size
Dim Size_Title As New com.sun.star.awt.Size
Dim Size_SubTitle As New com.sun.star.awt.Size
Dim Size_xTitle As New com.sun.star.awt.Size
Dim Size_yTitle As New com.sun.star.awt.Size
Dim Size_Legend As New com.sun.star.awt.Size
oRange = thisComponent.getCurrentSelection.getRangeAddress
oRect.Width = 34000
oRect.Height = 19500
oRect.X = 8650
oRect.Y = 20
cTitle = "AdminChart"
oCharts.addNewByName(cTitle,oRect,oRangeAddress(), TRUE, TRUE)
oChart = oCharts.getByName(cTitle).embeddedObject
oChart.Diagram = oChart.createInstance("com.sun.star.chart.LineDiagram") 'LineDiagram
oDiagram = oChart.Diagram
'Change position and size of chart area in rectangle
oChart.RefreshAddInAllowed = True
Pos_Chart.X = 700
Pos_Chart.Y = 1600
Size_Chart.Width = oRect.Width - 1500
Size_Chart.Height = oRect.Height - 2500
oDiagram.setPosition( Pos_Chart )
oDiagram.setSize( Size_Chart )
'Title and Subtitle of Chart
'oChart.SplineType = 0
oChart.HasMainTitle = True
oChart.Title.String = cTitle
oChart.Title.CharColor = cBlack
oChart.Title.CharFontName = "Liberation Sans"
oChart.Title.CharHeight = 16 ' Font Size
oChart.Title.CharWeight = 0 ' Bold in %
oChart.Title.CharPosture = 0 ' Italics = 1
oChart.HasSubTitle = False
'oChart.SubTitle.String = "Testing the waters"
'oChart.SubTitle.CharColor = cBlue
'oChart.SubTitle.CharFontName = "Liberation Sans"
'oChart.SubTitle.CharHeight = 12 ' Font Size
'oChart.SubTitle.CharWeight = 100 ' Bold in %
'oChart.SubTitle.CharPosture = 0 ' Italics
'Chart Area colours
oDiagram.Wall.FillStyle = com.sun.star.drawing.FillStyle.SOLID
oDiagram.Wall.FillColor = cWhite 'Chart Area Colour
oDiagram.Wall.LineStyle = com.sun.star.drawing.FillStyle.SOLID
oDiagram.Wall.LineColor = cBlack 'Chart Area Border Colour
oDiagram.Wall.LineWidth = 20
oChart.Area.FillStyle = com.sun.star.drawing.FillStyle.SOLID
oChart.Area.FillColor = cWhite 'Full Area Colour
'Horizontal Major Grid Lines
oDiagram.HasYAxisGrid = True
oDiagram.YMainGrid.LineStyle = com.sun.star.drawing.FillStyle.SOLID
oDiagram.YMainGrid.LineColor = cGray02
oDiagram.YMainGrid.LineWidth = 20
'Vertical Major Grid Lines
oDiagram.HasXAxisGrid = False
oDiagram.XMainGrid.LineStyle = com.sun.star.drawing.FillStyle.SOLID
oDiagram.XMainGrid.LineColor = cGray02
oDiagram.XMainGrid.LineWidth = 20
'X-Axis
oDiagram.HasXAxis = True
oDiagram.HasXAxisDescription = True
oDiagram.HasXAxisTitle = True
oXaxis = oDiagram.getXAxis()
oXaxis.AxisTitle.String = "X-Values"
oXaxis.AxisTitle.CharFontName = "Liberation Sans"
oXaxis.AxisTitle.CharColor = cBlack
oXaxis.AxisTitle.CharHeight = 11 ' Font Size
oXaxis.AxisTitle.CharWeight = 0 ' Bold in %
oXaxis.AxisTitle.CharPosture = 0 ' Italics = 1
oXaxis.AutoMin = True
oXaxis.AutoMax = True
'oXaxis.Min = 1
'oXaxis.Max = 5
oXaxis.CharColor = cBlack
oXaxis.CharFontName = "Liberation Sans"
oXaxis.CharHeight = 10 ' Font Size
oXaxis.CharWeight = 100 ' Bold in %
oXaxis.CharPosture = 0 ' Italics
oXaxis.LineColor = cBlack
oXaxis.LineWidth = 20
'Primary Y-Axis
oDiagram.HasYAxis = True
oDiagram.HasYAxisDescription = True
oDiagram.HasYAxisTitle = False
oDiagram.HasYAxis = True
oYaxis = oDiagram.getYAxis()
oYaxis.AxisTitle.String = "Y-Values"
oYaxis.AxisTitle.CharFontName = "Liberation Sans"
oYaxis.AxisTitle.CharColor = cBlack
oYaxis.AxisTitle.CharHeight = 11 ' Font Size
oYaxis.AxisTitle.CharWeight = 0 ' Bold in %
oYaxis.AxisTitle.CharPosture = 0 ' Italics = 1
oYaxis.AutoMin = False
oYaxis.AutoMax = False
oYaxis.Min = 0
oYaxis.Max = 2000
oYaxis.StepMain = 250
oYaxis.CharColor = cBlack
oYaxis.CharFontName = "Liberation Sans"
oYaxis.CharHeight = 10 ' Font Size
oYaxis.CharWeight = 0 ' Bold in %
oYaxis.CharPosture = 0 ' Italics
oYaxis.LineColor = cBlack
oYaxis.LineWidth = 20
oYaxis.LinkNumberFormatToSource = False
oYaxis.NumberFormat = "114"
oChart.HasLegend = 1
oLegend = oChart.getLegend()
oLegend.AutomaticPosition = True
oLegend.Alignment = com.sun.star.chart.ChartLegendPosition.BOTTOM
oLegend.CharHeight = 10 ' Font Size
oLegend.CharWeight = 0 ' Bold in %
oLegend.CharFontName = "Liberation Sans"
oLegend.FillColor = cWhite
oLegend.LineColor = cWhite
oLegend.CharColor = cBlack
oLegend.CharPosture = 0 'Italics
oDiagram.Lines = True
oDiagram.LineColor = cDarkRed
oDiagram.LineWidth = 20
'oChart.DataSourceLabelsInFirstColumn = True
'oChart.DataSourceLabelsInFirstRow = False
'Call Chart Procedure
If ThisComponent.NamedRanges.getByName("ChartType").getReferredCells.String = "Opbouw Stand van Zaken" Then
cTitle = "Opbouw Eigen Woning " & ThisComponent.NamedRanges.getByName("ChartPeriod").getReferredCells.String
Call ChartBuildStatus(xStartRow, xEndRow, cTitle)
End If
End Sub
Once the macro ChartBuildStatus is called, that macro contains the following actions:
I change the chart type of AdminChart to AreaDiagram. That shows up correctly on my screen.
I then try to make it a stacked diagram. But it won’t…
After that, I change some formatting characteristics of the chart, like axis titles and number formats
And I generate the category and data series calling on two other macro’s, CreateDataSequence() and CreateDataSeries_Chart(). I don’think they are the problem, they work correctly with other non-stacked line charts and the series
I have tried a number of things, including moving the stacking code to the end of the macro. But I cannot get it to work correctly. I must be overlooking something or doing something very stupid. So after literally days of trying I am pinning my hopes on the community. Anybody who can shed some light on this?
Sub ChartBuildStatus(xStartRow As Long, xEndRow As Long, cTitle As String)
'Define Colours
cBlack = RGB(0, 0, 0)
cGray01 = RGB(50, 50, 50)
cGray02 = RGB(191, 191, 191)
cWhite = RGB(255, 255, 255)
wshC = ThisComponent.Sheets.getByName("Charts")
wshM = ThisComponent.Sheets.getByName("Months")
oCharts = wshC.Charts
oChart = oCharts.getByName("AdminChart").embeddedObject
oChart.Diagram = oChart.createInstance("com.sun.star.chart.AreaDiagram")
'Stackable Diagram
oChart.Diagram = oChart.createInstance("com.sun.star.chart.StackableDiagram")
oDiagram = oChart.getDiagram()
oChart.setDiagram(oDiagram)
oDiagram = oChart.getDiagram()
With oDiagram
.Stacked = True
.Percent = False
.Vertical = False
End With
ChartUnit = 50000
'Specific Chart Adjustments
oXaxis = oDiagram.getXAxis()
oYaxis = oDiagram.getYAxis()
'Ymax And Ymin
oYaxis.AutoMax = True
oYaxis.Min = 0
'Chart details
oChart.Title.String = cTitle
oDiagram.HasXAxisTitle = False
oYaxis.AxisTitle.String = "EUR"
oYaxis.StepMain = ChartUnit
oYaxis.NumberFormat = "115" ' #,##0
'Data Series Generator
oDataProvider = oChart.getDataProvider()
oDiagram = oChart.getFirstDiagram()
oCooSys = oDiagram.getCoordinateSystems()
oCoods = oCooSys(0) ' this chart has only a coordinate system
oChartTypes = oCoods.getChartTypes() ' chart type one by one
oChartType = oChartTypes(0)
'Data Ranges
xEndRow = getLastContentIndex ("Months", 3 )
Xrange = "Months.C" & (xStartRow + 1) & ":C" & (xEndRow + 1) 'yyyymm
Y1range = "Months.S" & (xStartRow + 1) & ":S" & (xEndRow + 1) 'Overwaarde
Y2range = "Months.T" & (xStartRow + 1) & ":T" & (xEndRow + 1) 'Eigen Geld
Y3range = "Months.M" & (xStartRow + 1) & ":M" & (xEndRow + 1) 'Reguliere Aflossing
Y4range = "Months.N" & (xStartRow + 1) & ":N" & (xEndRow + 1) 'Extra Aflossing
Y5range = "Months.Q" & (xStartRow + 1) & ":Q" & (xEndRow + 1) 'Restant Hypotheek
'Prepare for Data Series
oDataProvider = oChart.getDataProvider()
oDiagram = oChart.getFirstDiagram()
oCooSys = oDiagram.getCoordinateSystems()
oCoods = oCooSys(0) ' this chart has only a coordinate system
oChartTypes = oCoods.getChartTypes() ' chart type one by one
oChartType = oChartTypes(0)
'X-Axis Data series
dim categorySequence as object, categoryRange as string
dim coordinateSystem as object, axis as object
dim scaleData as new com.sun.star.chart2.ScaleData
categorySequence = CreateUnoService("com.sun.star.chart2.data.LabeledDataSequence")
categorySequence.setValues(CreateDataSequence(oDataProvider, Xrange, "categories"))
coordinateSystem = oDiagram.getCoordinateSystems()(0)
axis = coordinateSystem.getAxisByDimension(0, 0)
scaleData = axis.getScaleData()
scaleData.Categories = categorySequence
axis.setScaleData(scaleData)
'Y-Axis Data series
oDataSeriesList = oChartType.getDataSeries()
Dim oNewDataSeriesList(4) As Object ' new data series
oSeries1 = CreateDataSeries_Chart(oDataProvider, Xrange, Y1range, "Overwaarde")
oSeries1.Color = RGB(197, 224, 180)
oSeries1.LineWidth = 60
oNewDataSeriesList(0) = oSeries1
oSeries2 = CreateDataSeries_Chart(oDataProvider, Xrange, Y2range, "EigenGeld")
oSeries2.Color = RGB(0, 176, 80)
oSeries2.LineWidth = 60
oNewDataSeriesList(1) = oSeries2
oSeries3 = CreateDataSeries_Chart(oDataProvider, Xrange, Y3range, "Aflossing")
oSeries3.Color = RGB(146, 208, 80)
oSeries3.LineWidth = 60
oNewDataSeriesList(2) = oSeries3
oSeries4 = CreateDataSeries_Chart(oDataProvider, Xrange, Y4range, "AflossingExtra")
oSeries4.Color = RGB(255, 215, 0)
oSeries4.LineWidth = 60
oNewDataSeriesList(3) = oSeries4
oSeries5 = CreateDataSeries_Chart(oDataProvider, Xrange, Y5range, "RestHypotheek")
oSeries5.Color = RGB(192, 0, 0)
oSeries5.LineWidth = 60
oNewDataSeriesList(4) = oSeries5
'Update chart (only the charttype is updated)
oChartType.setDataSeries(oNewDataSeriesList)
End Sub
On the left the chart as it looks now in my LibreOffice Calc spreadsheet, on the right the chart as I want it to look based on my current Excel sheet.
Left: LibreOffice / Right: Excel (how I want it)
For the sake of completeness, please find the additional macro’s CreateDataSequence() and CreateDataSeries_Chart() below.
CreateDataSequence()
Function CreateDataSequence( oDataProvider As Object, sRangeRepresentation As String, sRole As String ) As Object
Dim oDataSequence As Object
On Error GoTo Handler
' create .chart2.data.DataSequence from range representation
oDataSequence = oDataProvider.createDataSequenceByRangeRepresentation(sRangeRepresentation)
If NOT IsNull(oDataSequence) Then
oDataSequence.Role = sRole
End If
Handler:
CreateDataSequence = oDataSequence
End Function
CreateDataSeries_Chart()
Function CreateDataSeries_Chart( oDataProvider As Object, sXRangeRepresentation As String, sYRangeRepresentation As String, sLabelRangeRepresentation As String ) As Object
Dim oNewDataSeries As Object
oNewDataSeries = CreateUnoService("com.sun.star.chart2.DataSeries")
Dim oData(1) As Object ' x and y: .chart2.data.XLabeledDataSequence
' Y
oDataY = CreateUnoService("com.sun.star.chart2.data.LabeledDataSequence")
oSequenceY = CreateDataSequence(oDataProvider, sYRangeRepresentation, "values-y")
If NOT IsNull(oSequenceY) Then
oDataY.setValues(oSequenceY)
If NOT ((IsMissing(sLabelRangeRepresentation)) AND (sLabelRangeRepresentation <> "")) Then
oSequenceLabel = CreateDataSequence(oDataProvider, sLabelRangeRepresentation, "label") ' ""
oDataY.setLabel(oSequenceLabel) ' oSequenceLabel label is used as name
End If
End If
' X
oDataX = CreateUnoService("com.sun.star.chart2.data.LabeledDataSequence")
oSequenceX = CreateDataSequence(oDataProvider, sXRangeRepresentation, "values-x")
If NOT IsNull(oSequenceX) Then
oDataX.setValues(oSequenceX)
End If
' set x and y data to series
aData = Array(oDataY, oDataX)
oNewDataSeries.setData(aData)
CreateDataSeries_Chart = oNewDataSeries
End Function
I have 3 VGG: VGGA, VGGB and VGG*, trained with the following training function:
def train(nets, loaders, optimizer, criterion, epochs=20, dev=None, save_param=False, model_name="valerio"):
# try:
nets = [n.to(dev) for n in nets]
model_a = module_unwrap(nets[0], True)
model_b = module_unwrap(nets[1], True)
model_c = module_unwrap(nets[2], True)
reg_loss = nn.MSELoss()
criterion.to(dev)
reg_loss.to(dev)
# Initialize history
history_loss = {"train": [], "val": [], "test": []}
history_accuracy = {"train": [], "val": [], "test": []}
# Store the best val accuracy
best_val_accuracy = 0
# Process each epoch
for epoch in range(epochs):
# Initialize epoch variables
sum_loss = {"train": 0, "val": 0, "test": 0}
sum_accuracy = {"train": [0,0,0], "val": [0,0,0], "test": [0,0,0]}
progbar = None
# Process each split
for split in ["train", "val", "test"]:
if split == "train":
for n in nets:
n.train()
widgets = [
' [', pb.Timer(), '] ',
pb.Bar(),
' [', pb.ETA(), '] ', pb.Variable('ta','[Train Acc: {formatted_value}]')
]
progbar = pb.ProgressBar(max_value=len(loaders[split][0]),widgets=widgets,redirect_stdout=True)
else:
for n in nets:
n.eval()
# Process each batch
for j,((input_a, labels_a),(input_b, labels_b)) in enumerate(zip(loaders[split][0],loaders[split][1])):
input_a = input_a.to(dev)
input_b = input_b.to(dev)
labels_a = labels_a.long().to(dev)
labels_b = labels_b.long().to(dev)
#print(labels_a.shape)
#labels_a = labels_a.squeeze()
#labels_b = labels_b.squeeze()
#labels_a = labels_a.unsqueeze(1)
#labels_b = labels_b.unsqueeze(1)
#print(labels_a.shape)
#labels_a = labels_a.argmax(-1)
#labels_b = labels_b.argmax(-1)
inputs = torch.cat([input_a,input_b],axis=0)
labels = torch.cat([labels_a, labels_b])
#labels = labels.squeeze()
#print(labels.shape)
#labels = labels.argmax(-1)
# Reset gradients
optimizer.zero_grad()
# Compute output
features_a = nets[0](input_a)
features_b = nets[1](input_b)
features_c = nets[2](inputs)
pred_a = torch.squeeze(nets[3](features_a))
pred_b = torch.squeeze(nets[3](features_b))
pred_c = torch.squeeze(nets[3](features_c))
loss = criterion(pred_a, labels_a) + criterion(pred_b, labels_b) + criterion(pred_c, labels)
for n in model_a:
layer_a = model_a[n]
layer_b = model_b[n]
layer_c = model_c[n]
if (isinstance(layer_a,nn.Conv2d)):
loss += lambda_reg * reg_loss(combo_fn(layer_a.weight,layer_b.weight),layer_c.weight)
if (layer_a.bias is not None):
loss += lambda_reg * reg_loss(combo_fn(layer_a.bias, layer_b.bias), layer_c.bias)
# Update loss
sum_loss[split] += loss.item()
# Check parameter update
if split == "train":
# Compute gradients
loss.backward()
# Optimize
optimizer.step()
# Compute accuracy
#https://discuss.pytorch.org/t/bcewithlogitsloss-and-model-accuracy-calculation/59293/ 2
#pred_labels_a = (pred_a >= 0.0).long() # Binarize predictions to 0 and 1
#pred_labels_b = (pred_b >= 0.0).long() # Binarize predictions to 0 and 1
#pred_labels_c = (pred_c >= 0.0).long() # Binarize predictions to 0 and 1
#print(pred_a.shape)
_,pred_label_a = torch.max(pred_a, dim = 1)
pred_labels_a = (pred_label_a == labels_a).float()
_,pred_label_b = torch.max(pred_b, dim = 1)
pred_labels_b = (pred_label_b == labels_b).float()
_,pred_label_c = torch.max(pred_c, dim = 1)
pred_labels_c = (pred_label_c == labels).float()
batch_accuracy_a = pred_labels_a.sum().item() / len(labels_a)
batch_accuracy_b = pred_labels_b.sum().item() / len(labels_b)
batch_accuracy_c = pred_labels_c.sum().item() / len(labels)
# Update accuracy
sum_accuracy[split][0] += batch_accuracy_a
sum_accuracy[split][1] += batch_accuracy_b
sum_accuracy[split][2] += batch_accuracy_c
if (split=='train'):
progbar.update(j, ta=batch_accuracy_c)
if (progbar is not None):
progbar.finish()
# Compute epoch loss/accuracy
epoch_loss = {split: sum_loss[split] / len(loaders[split][0]) for split in ["train", "val", "test"]}
epoch_accuracy = {split: [sum_accuracy[split][i] / len(loaders[split][0]) for i in range(len(sum_accuracy[split])) ] for split in ["train", "val", "test"]}
# # Store params at the best validation accuracy
# if save_param and epoch_accuracy["val"] > best_val_accuracy:
# # torch.save(net.state_dict(), f"{net.__class__.__name__}_best_val.pth")
# torch.save(net.state_dict(), f"{model_name}_best_val.pth")
# best_val_accuracy = epoch_accuracy["val"]
print(f"Epoch {epoch + 1}:")
# Update history
for split in ["train", "val", "test"]:
history_loss[split].append(epoch_loss[split])
history_accuracy[split].append(epoch_accuracy[split])
# Print info
print(f"\t{split}\tLoss: {epoch_loss[split]:0.5}\tVGG 1:{epoch_accuracy[split][0]:0.5}"
f"\tVGG 2:{epoch_accuracy[split][1]:0.5}\tVGG *:{epoch_accuracy[split][2]:0.5}")
if save_param:
torch.save({'vgg_a':nets[0].state_dict(),'vgg_b':nets[1].state_dict(),'vgg_star':nets[2].state_dict(),'classifier':nets[3].state_dict()},f'{model_name}.pth')
For each epoch of training the result is this:
Then, I have a combined model which sums the weights of VGGA and VGGB:
DO = 'TEST'
if (DO=='TRAIN'):
train(nets, loaders, optimizer, criterion, epochs=50, dev=dev,save_param=True)
else:
state_dicts = torch.load('valerio.pth')
model1.load_state_dict(state_dicts['vgg_a']) #questi state_dict vengono dalla funzione di training
model2.load_state_dict(state_dicts['vgg_b'])
model3.load_state_dict(state_dicts['vgg_star'])
classifier.load_state_dict(state_dicts['classifier'])
test(model1,classifier,test_loader_all)
test(model2, classifier, test_loader_all)
test(model3, classifier, test_loader_all)
summed_state_dict = OrderedDict()
for key in state_dicts['vgg_star']:
if key.find('conv') >=0:
print(key)
summed_state_dict[key] = combo_fn(state_dicts['vgg_a'][key],state_dicts['vgg_b'][key])
else:
summed_state_dict[key] = state_dicts['vgg_star'][key]
model3.load_state_dict(summed_state_dict)
test(model3, classifier, test_loader_all)
where the test function is this:
def test(net,classifier, loader):
net.to(dev)
classifier.to(dev)
net.eval()
sum_accuracy = 0
# Process each batch
for j, (input, labels) in enumerate(loader):
input = input.to(dev)
labels = labels.float().to(dev)
features = net(input)
pred = torch.squeeze(classifier(features))
# https://discuss.pytorch.org/t/bcewithlogitsloss-and-model-accuracy-calculation/59293/ 2
#pred_labels = (pred >= 0.0).long() # Binarize predictions to 0 and 1
_,pred_label = torch.max(pred, dim = 1)
pred_labels = (pred_label == labels).float()
batch_accuracy = pred_labels.sum().item() / len(labels)
# Update accuracy
sum_accuracy += batch_accuracy
epoch_accuracy = sum_accuracy / len(loader)
print(f"Accuracy after sum: {epoch_accuracy:0.5}")
And the result of this aggregation is the following:
I want to modify my training function in order to print the same things of the first image, plus the accuracy of the aggregated model (the highlighted part in red of the second picture). So basically, for each epoch, accuracies of VGGA, VGGB, VGG* and combined VGG, print these accuracies and continue with the training. I tried to add this model combo but I failed, because I did not able to insert into each epoch, but only at the end of the training. I was trying to add in the training function, between print(f"Epoch {epoch + 1}:")and
# Update history
for split in ["train", "val", "test"]:
the code with the part of state_dict, but i am doing something wrong, i do not know what.
Can I reuse the code of the test, or I have to write new code?
Do you think i have to save the state_dict for each epoch, or i can do something else? Like model_c.parameters()=model_a.parameters()+model_b.parameters() (which does not work, already tried)
I solved, here is the solution of how I modified my training function:
def train(nets, loaders, optimizer, criterion, epochs=20, dev=None, save_param=False, model_name="valerio"):
# try:
nets = [n.to(dev) for n in nets]
model_a = module_unwrap(nets[0], True)
model_b = module_unwrap(nets[1], True)
model_c = module_unwrap(nets[2], True)
reg_loss = nn.MSELoss()
criterion.to(dev)
reg_loss.to(dev)
# Initialize history
history_loss = {"train": [], "val": [], "test": []}
history_accuracy = {"train": [], "val": [], "test": []}
history_test = 0
# Store the best val accuracy
best_val_accuracy = 0
# Process each epoch
for epoch in range(epochs):
# Initialize epoch variables
sum_loss = {"train": 0, "val": 0, "test": 0}
sum_accuracy = {"train": [0,0,0], "val": [0,0,0], "test": [0,0,0]}
progbar = None
# Process each split
for split in ["train", "val", "test"]:
if split == "train":
for n in nets:
n.train()
widgets = [
' [', pb.Timer(), '] ',
pb.Bar(),
' [', pb.ETA(), '] ', pb.Variable('ta','[Train Acc: {formatted_value}]')
]
progbar = pb.ProgressBar(max_value=len(loaders[split][0]),widgets=widgets,redirect_stdout=True)
else:
for n in nets:
n.eval()
# Process each batch
for j,((input_a, labels_a),(input_b, labels_b)) in enumerate(zip(loaders[split][0],loaders[split][1])):
input_a = input_a.to(dev)
input_b = input_b.to(dev)
labels_a = labels_a.long().to(dev)
labels_b = labels_b.long().to(dev)
#print(labels_a.shape)
#labels_a = labels_a.squeeze()
#labels_b = labels_b.squeeze()
#labels_a = labels_a.unsqueeze(1)
#labels_b = labels_b.unsqueeze(1)
#print(labels_a.shape)
#labels_a = labels_a.argmax(-1)
#labels_b = labels_b.argmax(-1)
inputs = torch.cat([input_a,input_b],axis=0)
labels = torch.cat([labels_a, labels_b])
#labels = labels.squeeze()
#print(labels.shape)
#labels = labels.argmax(-1)
# Reset gradients
optimizer.zero_grad()
# Compute output
features_a = nets[0](input_a)
features_b = nets[1](input_b)
features_c = nets[2](inputs)
pred_a = torch.squeeze(nets[3](features_a))
pred_b = torch.squeeze(nets[3](features_b))
pred_c = torch.squeeze(nets[3](features_c))
loss = criterion(pred_a, labels_a) + criterion(pred_b, labels_b) + criterion(pred_c, labels)
for n in model_a:
layer_a = model_a[n]
layer_b = model_b[n]
layer_c = model_c[n]
if (isinstance(layer_a,nn.Conv2d)):
loss += lambda_reg * reg_loss(combo_fn(layer_a.weight,layer_b.weight),layer_c.weight)
if (layer_a.bias is not None):
loss += lambda_reg * reg_loss(combo_fn(layer_a.bias, layer_b.bias), layer_c.bias)
# Update loss
sum_loss[split] += loss.item()
# Check parameter update
if split == "train":
# Compute gradients
loss.backward()
# Optimize
optimizer.step()
# Compute accuracy
#https://discuss.pytorch.org/t/bcewithlogitsloss-and-model-accuracy-calculation/59293/ 2
#pred_labels_a = (pred_a >= 0.0).long() # Binarize predictions to 0 and 1
#pred_labels_b = (pred_b >= 0.0).long() # Binarize predictions to 0 and 1
#pred_labels_c = (pred_c >= 0.0).long() # Binarize predictions to 0 and 1
#print(pred_a.shape)
_,pred_label_a = torch.max(pred_a, dim = 1)
pred_labels_a = (pred_label_a == labels_a).float()
_,pred_label_b = torch.max(pred_b, dim = 1)
pred_labels_b = (pred_label_b == labels_b).float()
_,pred_label_c = torch.max(pred_c, dim = 1)
pred_labels_c = (pred_label_c == labels).float()
batch_accuracy_a = pred_labels_a.sum().item() / len(labels_a)
batch_accuracy_b = pred_labels_b.sum().item() / len(labels_b)
batch_accuracy_c = pred_labels_c.sum().item() / len(labels)
# Update accuracy
sum_accuracy[split][0] += batch_accuracy_a
sum_accuracy[split][1] += batch_accuracy_b
sum_accuracy[split][2] += batch_accuracy_c
if (split=='train'):
progbar.update(j, ta=batch_accuracy_c)
if (progbar is not None):
progbar.finish()
# Compute epoch loss/accuracy
epoch_loss = {split: sum_loss[split] / len(loaders[split][0]) for split in ["train", "val", "test"]}
epoch_accuracy = {split: [sum_accuracy[split][i] / len(loaders[split][0]) for i in range(len(sum_accuracy[split])) ] for split in ["train", "val", "test"]}
# # Store params at the best validation accuracy
# if save_param and epoch_accuracy["val"] > best_val_accuracy:
# # torch.save(net.state_dict(), f"{net.__class__.__name__}_best_val.pth")
# torch.save(net.state_dict(), f"{model_name}_best_val.pth")
# best_val_accuracy = epoch_accuracy["val"]
print(f"Epoch {epoch + 1}:")
# Update history
for split in ["train", "val", "test"]:
history_loss[split].append(epoch_loss[split])
history_accuracy[split].append(epoch_accuracy[split])
# Print info
print(f"\t{split}\tLoss: {epoch_loss[split]:0.5}\tVGG 1:{epoch_accuracy[split][0]:0.5}"
f"\tVGG 2:{epoch_accuracy[split][1]:0.5}\tVGG *:{epoch_accuracy[split][2]:0.5}")
if save_param:
torch.save({'vgg_a':nets[0].state_dict(),'vgg_b':nets[1].state_dict(),'vgg_star':nets[2].state_dict(),'classifier':nets[3].state_dict()},f'{model_name}.pth')
test(nets[0], nets[3], test_loader_all)
test(nets[1], nets[3], test_loader_all)
test(nets[2], nets[3], test_loader_all)
summed_state_dict = OrderedDict()
for key in nets[2].state_dict():
if key.find('conv') >=0:
#print(key)
summed_state_dict[key] = combo_fn(nets[0].state_dict()[key],nets[1].state_dict()[key])
else:
summed_state_dict[key] = nets[2].state_dict()[key]
nets[2].load_state_dict(summed_state_dict)
test(nets[2], nets[3], test_loader_all)
The edited parts are the last rows.
I am trying to have a matrix which its elements doesn't have the same size .
Say the element_1 = 0.1234567 and the element_2 = 0.1 and I need the element_2 =0.1000000 so that both of them has the same size.
clc;clear all
a = rand(4,12);
COL_Names ={'This_is_Colu_No_1','This_is_Colu_No_2','This_is_Colu_No_3','This_is_Colu_No_4','This_is_Colu_No_5','This_is_Colu_No_6','This_is_Colu_No_7','This_is_Colu_No_8','This_is_Colu_No_9','This_is_Colu_No_10','This_is_Colu_No_11','This_is_Colu_No_12'};
rowNames = {'ROW1';'ROW2';'ROW3';'ROW4'};
T = array2table(a,'VariableNames',COL_Names,'RowNames',rowNames);
writetable(T,'Data.txt','Delimiter','\t','WriteRowNames',true);
type Data.txt ;
The OutPut is like this
Row This_is_Colu_No_1 This_is_Colu_No_2 This_is_Colu_No_3 This_is_Colu_No_4 This_is_Colu_No_5 This_is_Colu_No_6 This_is_Colu_No_7 This_is_Colu_No_8 This_is_Colu_No_9 This_is_Colu_No_10 This_is_Colu_No_11 This_is_Colu_No_12
ROW1 0.139740979774291 0.231035232035157 0.347778782863186 0.279682446566279 0.060995054119542 0.233212699943628 0.507599581908539 0.833087779293817 0.552819386888535 0.43251811668393 0.342580158122272 0.420574544492339
ROW2 0.00459708931895875 0.703626845695885 0.33064632159971 0.85782393462353 0 0.935097755896966 0.582441521353621 0.155241648807001 0.163717355897126 0.48985529896707 0.0134551766978835 0.810989133317225
ROW3 0.791254563282513 0.650747335567064 0.293769172888192 0.15110222627643 0.962791661993452 0.842147123142386 0.586462512126695 0.109349751268813 1 0.00525695361457879 0.700826048054212 0.989915984093474
ROW4 0.513993416249574 0.868158891144176 0.293769172888 0.552496163682282 0.301098948730568 0.779790450269442 0.420527994140777 0.523231514251179 0.0602548802340035 0.261436547849062 0.84923648156472 0.433189006269314
I think you need to do it manually using fprintf with formatSpec:
clc, clear, rng(3);
a = rand(4, 3);
colNames = {'This_is_Colu_No_9', 'This_is_Colu_No_10', 'This_is_Colu_No_11'};
rowNames = {'ROW98'; 'ROW99'; 'ROW100'; 'ROW101'};
formatSpecHead = '%-6s %-22s %-22s %-22s\n';
formatSpecRow = '%-6s %.20f %.20f %.20f\n';
fid = fopen('a.fwf', 'w');
fprintf(fid, formatSpecHead, 'Row', colNames{:}); % write header
for row = 1:size(a, 1)
fprintf(fid, formatSpecRow, rowNames{row}, a(row, :)); % write row
end
fclose(fid);
Then a.fwf looks like:
Row This_is_Colu_No_9 This_is_Colu_No_10 This_is_Colu_No_11
ROW98 0.55079790257457550418 0.89294695434765469777 0.05146720330082987793
ROW99 0.70814782261810482744 0.89629308893343806464 0.44080984365063646813
ROW100 0.29090473891294432729 0.12558531046383625274 0.02987621087856695556
ROW101 0.51082760519766301499 0.20724287813818675907 0.45683322439471107934
I have an Excel form which reads data from each row of an 80-row worksheet and populates each cell into a textbox.
The code for the initial form is below.
Private Sub UserForm_Initialize()
txtbox_revri_idnum.Text = Worksheets("Risk&Issues").Range("A4").Value
txtbox_revri_projname.Text = Worksheets("Risk&Issues").Range("B4").Value
txtbox_revri_isrefnum.Text = Worksheets("Risk&Issues").Range("C4").Value
txtbox_revri_riskrefnum.Text = Worksheets("Risk&Issues").Range("D4").Value
...
txtbox_revri_projname.SetFocus
End Sub
Ideally, what I'd like to do is when the 'next' button is pressed, the next row (A5, B5, C5, D5...) is parsed into an array, and the Range is updated with the next row's data. Conversely, the 'previous' button would go in the reverse direction (A3, B3, C3, D3...). Essentially, I'd like to navigate up and down the worksheet reviewing the entire row with each cell in a new textbox.
I've tried all sorts of combinations of VBA (baring in mind I'm new to VBA although do have some Java and PHP experience). The code I have for my 'next' button current has the code below.
Private Sub button_revri_next_Click()
txtbox_revri_idnum.Text = ActiveCell.Next.Text
...
End Sub
Thank you... I know you guys are way more smarter than me!
You can use .Offset to achieve what you want.
TRIED AND TESTED
Dim i As Long, j As Long
Dim rng As Range
Private Sub UserForm_Initialize()
Set rng = Worksheets("Risk&Issues").Range("A4")
i = 0: j = 1
txtbox_revri_idnum.Text = rng.Offset(i).Value
txtbox_revri_projname.Text = rng.Offset(i, j).Value: j = j + 1
txtbox_revri_isrefnum.Text = rng.Offset(i, j).Value: j = j + 1
txtbox_revri_riskrefnum.Text = rng.Offset(i, j).Value: j = j + 1
'
'
txtbox_revri_projname.SetFocus
End Sub
'~~> Next Button
Private Sub button_revri_next_Click()
i = i + 1: j = 1
If i > (Sheets("Risk&Issues").Rows.Count - 4) Then
MsgBox "Max rows Reached"
Exit Sub
End If
txtbox_revri_idnum.Text = rng.Offset(i).Value
txtbox_revri_projname.Text = rng.Offset(i, j).Value: j = j + 1
txtbox_revri_isrefnum.Text = rng.Offset(i, j).Value: j = j + 1
txtbox_revri_riskrefnum.Text = rng.Offset(i, j).Value: j = j + 1
'
'
txtbox_revri_projname.SetFocus
End Sub
Similarly Previous Button
'~~> Previous Button
Private Sub button_revri_prev_Click()
i = i - 1: j = 1
If i < 0 Then
MsgBox "1st Row Reached"
Exit Sub
End If
txtbox_revri_idnum.Text = rng.Offset(i).Value
txtbox_revri_projname.Text = rng.Offset(i, j).Value: j = j + 1
txtbox_revri_isrefnum.Text = rng.Offset(i, j).Value: j = j + 1
txtbox_revri_riskrefnum.Text = rng.Offset(i, j).Value: j = j + 1
'
'
txtbox_revri_projname.SetFocus
End Sub
How can I vectorize this strsplit function in Matlab so that it runs faster? Thank you
function parts = strsplit(splitstr, str)
splitlen = length(splitstr);
k = strfind(str, splitstr);
parts = cell(length(k)+1, 1);
s = 1;
if isempty(k)
parts{1} = str;
return
end
for i=1:length(k)
parts{i} = str(s : k(i)-1);
s = k(i) + splitlen ;
end
parts{end} = str((k(end) + splitlen) : end);
You can replace your function by a call to TEXTSCAN.
str = 'testxyztest2xyztest3';
splitStr = 'xyz';
out = textscan(str,'%s','delimiter',splitStr,'multipleDelimsAsOne',1)
out{1}
ans =
'test'
'test2'
'test3'
You can use regexp (note also the example on string splitting on that page):
out = regexp(str, regexptranslate('escape', splitStr), 'split');
The regexptranslate function escapes the split string so that it gets treated as a literal string and not a regular expression.