Besides the default tabular format, multidimensional GAMS symbols can be visualized as graphs. GAMS MIRO offers comprehensive visualization options. A lot of plotting types are available and only need to be configured, i.e. adapted to your model-specific data.

Advanced: Symbol dimensions

If you want to configure a graphic for a GAMS symbol, you have to be aware of how many dimensions your GAMS symbol has, as this influences the type of graph to choose. GAMS symbols sometimes have many dimensions. Different chart types allow you to visualize a different number of dimensions. Whereas a pie chart allows you to visualize two dimensions - a numerical one that defines the area of the slices and a categorical one that defines the labels of the slices - a bar chart allows three dimensions to be visualized: a numerical dimension that defines the height of the bars, a categorical one that defines the discrete set of bars on the x-axis, and another categorical one that defines the color of the bars. In fact, we could even display a fourth, numerical dimension that defines the width of the bars.

As an example we can look at the parameter d(i,j) 'distance in thousands of miles' in the model transport. In addition to the domains 'i' - the set of plants - and 'j' - the set of markets - there is the value column, i.e. a total of 3 dimensions:

We just learned that bar charts allow us to visualize 3 dimensions, so let's choose a bar chart:

What if we had a fourth domain, e.g. another categorical dimension? A scatter plot allows us to visualize a forth dimension in the form of the marker type (e.g. circle, square, triangle etc.).

The more dimensions we try to plot in one chart, the harder it is to understand what is actually going on. Instead of trying to plot as many dimensions as possible in a single chart, we should think about reducing the number of dimensions. One way to do this is to animate our chart (e.g. to show how our data changes over time). Another way is to use a filter that allows us to select a particular slice of our data.

The following sections show some examples for different chart types you can use in MIRO. It is not useful to describe all possible options MIRO supports. Instead, you should just explore it yourself! Start the Configuration Mode and play around with the data until you get a satisfactory result.

The default renderer for all symbols is the MIRO pivot table. This powerful tool offers many useful features that allow you to explore your data interactively. Its basic functionality is to filter, aggregate and pivot data. In addition, the tool includes a variety of rendering capabilities that enable dynamic data visualization and analysis. Options range from a traditional table to visually appealing displays such as heat maps, line charts, bar charts, pie charts, scatter plots, time series, and more, making it easy to create charts that meet your analytical needs. Moreover, you can integrate multiple chart types within a single view, allowing for comprehensive data exploration and interpretation. With a good balance between server and client resource usage, it is designed to handle very large amounts of data.

The MIRO pivot table renderer allows you to store "views" - the current representation of the data - to the database and load it together with the scenario data. Simply click the button and assign a name to the view. In the dialog that opens, you can also give the view a title and, depending on the chart type, create axis labels, use logarithmic axes, and use mixed chart types. When you save the scenario, all your views are automatically stored as well. Load a view by first clicking on the "Load view" button and then selecting which view to load.

A click on the button toggles the presentation mode, which allows the pivot tool to look like a normal graphic renderer without pivot controls. The combination of predefined views and the presentation mode, which can be set as the default display, offers extensive possibilities for data visualization and in-depth analyses. A slice, or "view" of data can be exported as a PNG or CSV file.

Configuration

You can specify the default view as well as the default renderer (table/chart) of the pivot table renderer in the Configuration Mode. Note that unlike other charting tools, you configure most of the MIRO pivot table directly in the preview on the right. Drag and drop the domains where you want them to be and click on "Save". The symbol is loaded in MIRO with the same view that you have set up in the Configuration Mode.

Example: Model Transport, parameter shipment quantities in cases

For an example in GAMS, see the map example.

JSON example

{
  "dataRendering": {
    "schedule": {
      "outType": "miroPivot",
      "height": 700,
      "options": {
        "aggregationFunction": "sum",
        "pivotRenderer": "stackedbar",
        "enableHideEmptyCols": false,
        "hidePivotControls": true,
        "rows": "i",
        "filter": {
          "Hdr": "quantities"
        },
        "cols": {
          "j": null
        }
      }
    }
  }
}
                                    

Hide empty columns:
In addition to the configuration of the default view, there is also the possibility to activate a switch that allows to hide empty columns. This is especially useful if you use the MIRO Pivot Table for a data cube. A data cube is a collection of GAMS parameters that are combined to form a large cube. For example, if you have location information in the two parameters: ilocData(i,locHdr) and jlocData(j,locHdr), you can combine them into a cube with four dimensions: one dimension to store the name of the GAMS symbols: symname and the three dimensions i, j and locHdr. Note that the dimension i is missing in the parameter jlocData and vice versa. So we end up with a sparse cube:

You probably noticed that in this example missing dimensions in a symbol are filled with -. But we could have chosen any other UEL to indicate missing values. If you would visualize the data of this parameter with the MIRO Pivot Table, you might not want to see columns containing only -. Therefore, the MIRO Pivot Table has the options to hide these empty columns. The button to hide/show empty columns is located in the settings dialog.

Hide pivot controls (presentation mode):
This option activates the presentation mode by default. In this mode, all pivot controls are hidden so that the chart or table takes center stage. Only configured views and export buttons remain visible. The presentation mode can be very useful for users who do not want to play around with the data in the pivot tool, but only want to view predefined views/reports.

Enable fixed columns:
If this option is activated, the columns of the pivot table are fixed to the left side of the table and remain visible when scrolling horizontally. Note that this can cause display problems for very wide tables (e.g. with many columns), especially on small screens. For this reason, this option is automatically disabled for screens less than 768 pixels wide.

Use external default view:
If this option is enabled, a scenario-specific or global view is used as the default view for this symbol. If the specified view does not exist, the default configuration for this symbol is used.

MIRO uses Plotly to render pie charts.

Configuration

Any GAMS symbol with a value column can be displayed as a pie chart.

Example: Model Pickstock, parameter weight

Set         symbol           'stock symbol';
Variable    w(symbol)        'what part of the portfolio';

$onExternalOutput
Parameter   weight(symbol)   'weight of stock';
$offExternalOutput

stock_weight(s) = w.l(s);

In a resulting GDX this parameter looks as follows:

In the Configuration Mode we can then configure a pie chart.

Note:
Instead of the parameter weight, to which you assign the values of the variable w, you could have used a pie chart directly for the levels of variable w. However, this variable also contains all the zero values (due to nonzero upper bounds and scale factors) that would then become visible in the pie chart, see below. If you assign the levels of the variable w (w.l(symbol)) to the parameter weight instead, zeros will be squeezed out.

Multi-pie chart:
In a pie chart only one dimension, i.e. one column of a table, can be displayed. However, it is possible to display several pie charts for one symbol at the same time. This makes it possible to easily distribute a multidimensional symbol over several pie charts. This is comparable to e.g. a line chart, where new lines can be displayed by including new table headers.

JSON example

{
  "dataRendering": {
    "stock_weight": {
      "outType": "dtGraph",
      "graph": {
        "title": "Portfolio composition",
        "tool": "plotly",
        "type": "pie",
        "traces": {
          "1": {
            "labels": "symbol",
            "values": "value",
            "hole": 0,
            "name": "weight"
          }
        },
        "showlegend": true,
        "staticPlot": false
      },
      "height": 700
    }
  }
}
                                    

MIRO uses Plotly to render donut charts.

Configuration

A donut chart is a pie chart with a "hole". In the Configuration Mode the donut chart is therefore to be found under pie chart. In addition to a standard pie chart, only the size of the donut "hole" can be configured.

Example: Model Pickstock, parameter weight

JSON example

{
  "dataRendering": {
    "stock_weight": {
      "outType": "dtGraph",
      "graph": {
        "title": "Portfolio composition",
        "tool": "plotly",
        "type": "pie",
        "traces": {
          "1": {
            "labels": "symbol",
            "values": "value",
            "hole": 0.6,
            "name": "weight"
          }
        },
        "showlegend": true,
        "staticPlot": false
      },
      "height": 700
    }
  }
}
                                    

MIRO uses Plotly to render bar charts.

Configuration

If a symbol is to be visualized as a bar chart, an index must be specified for the x-axis as well as for the y-axis. Several data series can be plotted together on the x-axis.

Example: Model Pickstock, parameter weight

Set         symbol           'stock symbol';
Variable    w(symbol)        'what part of the portfolio';

$onExternalOutput
Parameter   weight(symbol)   'weight of stock';
$offExternalOutput

stock_weight(s) = w.l(s);

In a resulting GDX this parameter looks as follows:

In the Configuration Mode we then can configure a bar chart.

Note:
Instead of the parameter weight, to which you assign the values of the variable w, you could have used a bar chart directly for the levels of variable w. However, this variable also contains all the zero values that would then become visible in the chart, see the pie chart example.

JSON example

{
  "dataRendering": {
    "stock_weight": {
      "outType": "graph",
      "graph": {
        "title": "Portfolio composition",
        "tool": "plotly",
        "type": "bar",
        "barmode": "group",
        "ydata": {
          "value": {
            "label": "weight",
            "mode": "lines",
            "marker": {
              "line": {
                "width": 0
              },
              "color": "#ed0652"
            }
          }
        },
        "xdata": "symbol",
        "showlegend": false,
        "xaxis": {
          "title": "stock symbol",
          "showgrid": false,
          "zeroline": false,
          "showticklabels": true,
          "categoryorder": "trace"
        },
        "yaxis": {
          "title": "weight",
          "showgrid": false,
          "zeroline": false,
          "showticklabels": true,
          "categoryorder": "trace"
        }
      },
      "height": 700
    }
}
                                    

MIRO uses Plotly to render scatter plots.

Configuration

Example: Model Pickstock, parameter absolute error

The parameter to plot represents the absolute errors in the training phase.
We define a parameter abserror(td, 'absolute error train') (last row), to which we assign the values of the parameter error(d), but only those values where ord(d) <= trainingdays.

Set         date           'date';
Parameter   error(date)    'Absolute error';

td(d)     = ord(d) <= trainingdays;
ntd(d)    = not td(d);

solve pickStock min obj using mip;

error(d)  = abs(index(d)-fund(d));

Set errHdr    'stock symbol header'    / 'absolute error train', 'absolute error test' /;

$onExternalOutput
Table abserror(date,errHdr)          'absolute error';
$offExternalOutput

abserror(td, 'absolute error train')   = error(td);
abserror(ntd,'absolute error test')    = error(ntd);

In a resulting GDX this parameter looks as follows:

In the Configuration Mode we then can configure a scatter plot.

JSON example

{
  "dataRendering": {
    "abserror": {
      "outType": "graph",
      "graph": {
        "title": "Absolute error",
        "tool": "plotly",
        "type": "scatter",
        "ydata": {
          "absolute error train": {
            "label": "absolute error train",
            "mode": "markers",
            "fill": "none",
            "marker": {
              "symbol": "circle",
              "opacity": 1,
              "size": "12",
              "line": {
                "width": "2",
                "color": "#000000"
              },
              "color": "#ff0000"
            },
            "showlegend": false
          }
        },
        "xdata": "date",
        "showlegend": true,
        "xaxis": {
          "title": "date",
          "showgrid": false,
          "zeroline": false,
          "showticklabels": true,
          "categoryorder": "trace",
          "rangefrom": "2016-01-01",
          "rangeto": "2016-05-01"
        },
        "yaxis": {
          "title": "absolute error train",
          "showgrid": false,
          "zeroline": false,
          "showticklabels": true,
          "categoryorder": "trace"
        }
      },
      "height": 700
    }
  }
}
                                    

MIRO uses Plotly to render line charts.

Configuration

Line charts also include area charts.

Example: Model Pickstock, parameter absolute error

For an example in GAMS, see the scatter plot example.

JSON example

{
  "dataRendering": {
    "abserror": {
      "outType": "graph",
      "graph": {
        "title": "Absolute error",
        "tool": "plotly",
        "type": "scatter",
        "ydata": {
          "absolute error train": {
            "label": "absolute error train",
            "mode": "lines",
            "line": {
              "width": 2,
              "shape": "linear",
              "dash": "solid"
            },
            "showlegend": false,
            "fill": "tozeroy"
          },
          "absolute error test": {
            "label": "absolute error test",
            "mode": "lines",
            "line": {
              "width": 2,
              "shape": "linear",
              "dash": "solid"
            },
            "showlegend": false,
            "fill": "tozeroy"
          }
        },
        "xdata": "date",
        "showlegend": true,
        "xaxis": {
          "title": "date",
          "showgrid": false,
          "zeroline": false,
          "showticklabels": true,
          "categoryorder": "trace"
        },
        "yaxis": {
          "title": "absolute error train",
          "showgrid": false,
          "zeroline": false,
          "showticklabels": true,
          "categoryorder": "trace"
        },
        "plot_bgcolor": "rgba(95,95,95,0.08)"
      },
      "height": 700
    }
  }
}
                                    

MIRO uses Plotly to render bubble charts.

Configuration

A bubble chart is a scatter diagram with an additional dimension in the form of the size of the "bubbles".

Example: Model Pickstock, parameter absolute error

For an example in GAMS, see the scatter plot example.

JSON example

{
  "dataRendering": {
    "abserror": {
      "outType": "graph",
      "graph": {
        "title": "Absolute error",
        "tool": "plotly",
        "type": "bubble",
        "ydata": {
          "absolute error train": {
            "label": "absolute error train",
            "mode": "markers",
            "marker": {
              "symbol": "circle",
              "opacity": 1,
              "size": "absolute error train",
              "color": "absolute error train",
              "line": {
                "width": 0
              },
              "maxsize": 20
            },
            "showlegend": false
          }
        },
        "xdata": "date",
        "showlegend": false,
        "xaxis": {
          "title": "date",
          "showgrid": false,
          "zeroline": false,
          "showticklabels": true,
          "categoryorder": "trace",
          "rangefrom": "2016-01-01",
          "rangeto": "2016-05-01"
        },
        "yaxis": {
          "title": "absolute error train",
          "showgrid": false,
          "zeroline": false,
          "showticklabels": true,
          "categoryorder": "trace"
        }
      },
      "height": 700
    }
  }
}
                                    

MIRO uses Plotly to render histograms.

Configuration

Example: Model Pickstock, parameter absolute error

For an example in GAMS, see the scatter plot example.

JSON example

{
  "dataRendering": {
    "abserror": {
      "outType": "graph",
      "graph": {
        "title": "Absolute error",
        "tool": "plotly",
        "type": "hist",
        "xdata": {
          "absolute error train": {
            "labels": "absolute error train",
            "color": "rgba(40,181,35,0.73)"
          },
          "absolute error test": {
            "labels": "absolute error test",
            "color": "rgba(255,0,0,0.62)"
          }
        },
        "histnorm": "",
        "nbins": 5,
        "barmode": "overlay",
        "alpha": 0.6,
        "xaxis": {
          "title": "absolute error train"
        },
        "cumulative": false,
        "horizontal": false,
        "yaxis": {
          "title": "Frequency"
        },
        "showlegend": true
      },
      "height": 700
    }
  }
}
                                    

MIRO uses Leaflet to render maps.

Configuration

Example: Model Transport, parameter shipment quantities in cases

The parameter to plot represents the shipment quantities between canning plants and markets.

Set
    i           'canning plants'
    j           'markets'
    scheduleHdr 'schedule header' / 'lngP', 'latP', 'lngM', 'latM', 'cap', 'demand', 'quantities' /;

$onExternalOutput
Table  schedule(i,j,scheduleHdr) 'shipment quantities in cases';
$offExternalOutput

schedule(i,j, 'lngP')       = iLocData(i,'lng');
schedule(i,j, 'latP')       = iLocData(i,'lat');
schedule(i,j, 'lngM')       = jLocData(j,'lng');
schedule(i,j, 'latM')       = jLocData(j,'lat');
schedule(i,j, 'cap')        = a(i);
schedule(i,j, 'demand')     = b(j);
schedule(i,j, 'quantities') = x.l(i,j);

To be able to set markers on a map, the map tool requires latitude and longitude information. By declaring the parameter schedule as a table, the elements of the last index set scheduleHdr (lngP, latP, lngM, latM, cap, demand, quantities) can be used for the map configuration. You can find a detailed example here.

In a resulting GDX this parameter looks as follows:

In the Configuration Mode we then can configure a map.

$onExternalOutput
Parameter schedule(i,j,latP, lngP, latM, lngM) 'shipment quantities in cases';
$offExternalOutput

schedule('Seattle',  'New-York', '47.608013', '-122.335167', '40.730610', '-73.935242') = 50;
schedule('Seattle',  'Chicago',  '47.608013', '-122.335167', '41.881832', '-87.623177') = 300;
schedule('San-Diego','New-York', '32.715736', '-117.161087', '40.730610', '-73.935242') = 275;
schedule('San-Diego','Topeka',   '32.715736', '-117.161087', '39.056198', '-95.695312') = 275;

With regard to map-based graphics, a multitude of possible visualizations are conceivable. In the Configuration Mode of MIRO, the following features can be realized on a map:

• Markers
  Markers can be set. For this purpose, the position of the markers must be available in the data as longitude and latitude information (see table above). Marker labels can be set dynamically, i.e. based on data. In the following example, the marker label is set to Market: [j]. If you now click on a marker on the resulting map, the label will not be Market: [j]. Instead, the [j] is replaced with the element of the GAMS index set belonging to the marker (here: Chicago).
  
  The second index of the symbol schedule contains the markets j. The indices lngM and latM were selected as longitude and latitude data. For a given longitude-latitude combination, the corresponding element from the second index j is now displayed as label.
• Flows
  Flows can be displayed between different points on the map. If configured accordingly, the thickness of the arrows can depend on the flow data. The data overview from the table above also helps to understand this: In the example below the flows have been configured to point from lngP/latP to lngM/latM. The thickness depends on the data in the column quantities.
  Note:

  Flows cannot be configured for sets, since numeric values are required for the flow data.

  
  Unique flow labels:

  When configuring flows, you can optionally specify a unique label for each flow. This label appears when you click on a flow. It is important that the label is different for each flow. If this is not the case, only one flow is displayed for each duplicated label. A suitable configuration for a label is to specify the originator and the recipient for each flow. You can achieve this by using the square brackets syntax to address the symbol/data domains. In the example above the symbol schedule is used to display a map:

  Set
      i           'canning plants'
      j           'markets'
      scheduleHdr 'schedule header' / 'lngP', 'latP', 'lngM', 'latM', 'cap', 'demand', 'quantities' /;
  
  $onExternalOutput
  Table  schedule(i,j,scheduleHdr) 'shipment quantities in cases';
  $offExternalOutput
  

  Originator and recipient in this case are the domains i and j of the symbol schedule. If you configure the label as follows: From [i] to [j] - then the originator and recipient cities appear as labels for each arrow:

  
• Charts
  You can display small diagrams on your map. The following example shows a pie chart for each of the 16 German states.

  Example: Model Simple (see GAMS MIRO gallery), parameter energy mix report

  
  The graph is based on the data below:
  
  The position of each pie chart results from a longitude/latitude combination. The data of the pie chart for each state were determined in Configuration Mode with Renewable and Fossil.
  You also have the option to define a time dimension in the charts. The data of the example covers a period of one year with a resolution of one hour. The first column was defined as the time dimension (starting with t001. In this way, a separate pie chart is displayed for each period for each state (see play button at the bottom right of the map).
  Note:

  Charts on maps cannot be configured for sets, because numeric values are required for the chart data.

  Note that all data of your entire graph must come from the same GAMS symbol. If you want to have more sophisticated graphics, it may make sense to combine data from different symbols in one symbol. Or you can use our R API and create your own graphics.

Layers
For map visualizations it can be helpful to display different layers, e.g. with markers. In Configuration Mode you can specify such layers. In the resulting graphic you can select the different layers and display only the corresponding markers.

Example: Model Transport, parameter shipment quantities in cases

JSON example

{
  "dataRendering": {
    "schedule": {
      "outType": "graph",
      "graph": {
        "title": "shipment quantities in cases",
        "tool": "leaflet",
        "layersControl": {
          "options": {
            "collapsed": true
          },
          "position": "topright",
          "baseGroups": ["plants", "markets"]
        },
        "markers": {
          "1": {
            "lng": "lngp",
            "lat": "latp",
            "labelOptions": {
              "textsize": "12px",
              "permanent": false
            },
            "label": "Plant [i]",
            "group": "plants"
          },
          "2": {
            "lng": "lngm",
            "lat": "latm",
            "labelOptions": {
              "textsize": "12px",
              "permanent": false
            },
            "label": "Market [j]",
            "group": "markets"
          }
        },
        "flows": {
          "1": {
            "lng0": "lngp",
            "lat0": "latp",
            "lng1": "lngm",
            "lat1": "latm",
            "flow": "quantities",
            "color": "#0000ff",
            "minThickness": 1,
            "maxThickness": 10
          }
        }
      },
      "height": 700
    }
  }
}
                                    

MIRO uses DyGraphs to render time series diagrams.

Configuration

As the name of the graphic type suggests, this chart type is useful if you want to visualize time series data. An important property of such time series is the format of the data. You should make sure that the elements of the index set, which should represent the time component of your graphic, have their data available in a date format or at least can be formatted into such a format. A suitable syntax for a date would be yyyy-MM-dd, e.g. 2014-07-13. For a finer granularity, the format yyyy-MM-dd HH:mm:ss is recommended.

Example: Model Pickstock, parameter dow jones vs. index fund

Set date      'date'
    fHdr      'fund header'   / dj 'dow jones','index fund'  /;

$onExternalOutput
Table   dowVSindex(date,fHdr)  'dow jones vs. index fund';
$offExternalOutput

dowVSindex(d,'dj')             = index(d);
dowVSindex(d,'index fund')     = fund(d);

The index set date looks as follows:

The parameter dowVSindex:

In the Configuration Mode we then can configure a time series diagram:

The picture in this example shows a red dotted vertical line showing the "last day of training phase". A scalar output value has been integrated into the graph to define this event.
Note: To use such a scalar in the graphic, MIRO must of course know this symbol. You therefore need to tag it with $onExternalInput / $offExternalInput or $onExternalOutput / $offExternalOutput in the GAMS model.

The tool for creating time series diagrams supports the integration of static or dynamic scalar values. A static value could show a limit in the data in the form of a horizontal line. A dynamic scalar can be a date which results from the model calculations (i.e. output scalar) and is displayed in the graphic as an vertical event line.

The following options are available:

• Event line
  
• Limit line
  
• Annotation
  
• Shading
  

JSON example

{
  "dataRendering": {
    "dowvsindex": {
      "outType": "graph",
      "graph": {
        "title": "Dow Jones vs. Index fund",
        "tool": "dygraphs",
        "xdata": "date",
        "ydata": {
          "dj": {
            "label": "dow jones",
            "stemPlot": false,
            "stepPlot": false,
            "fillGraph": false,
            "drawPoints": false,
            "pointShape": "dot",
            "pointSize": 2
          },
          "index fund": {
            "label": "index fund",
            "stemPlot": false,
            "stepPlot": false,
            "fillGraph": false,
            "drawPoints": false,
            "pointShape": "dot",
            "pointSize": 2
          }
        },
        "dyOptions": {
          "includeZero": false,
          "logscale": false,
          "drawGrid": true,
          "stepPlot": false,
          "stemPlot": false,
          "fillGraph": false,
          "fillAlpha": 0.15,
          "drawPoints": false,
          "pointShape": "dot",
          "pointSize": 2
        },
        "dyHighlight": {
          "highlightCircleSize": 3,
          "highlightSeriesBackgroundAlpha": 0.5,
          "hideOnMouseOut": true
        },
        "xaxis": {
          "title": "date"
        },
        "yaxis": {
          "title": "dow jones"
        },
        "dyRangeSelector": {
          "height": 30,
          "strokeColor": "#808fab",
          "fillColor": "#a7b1c4",
          "retainDateWindow": false,
          "keepMouseZoom": true
        },
        "dyEvent": {
          "lastdaytraining": {
            "labelLoc": "top",
            "color": "rgb(255,0,0)",
            "strokePattern": "dashed",
            "label": "last day of training phase"
          }
        }
      },
      "height": 700
    }
  }
}
                                    

MIRO uses timevis to render gantt charts.

Configuration

The GAMS symbol whose data is to be visualized as a Gantt chart must contain the following index sets:

• Start
  Each event must have a start date. The GAMS elements of this set must be in date format, e.g. yyyy-mm-dd hh:mm:ss (hh:mm:ss is not required).
• Content
  Each event must have a content which is then shown as a label. The GAMS elements of the underlying set can be of any format.

Optional index sets of the GAMS Symbol:

• End
  If an event is not a single point in time, but rather a range, you can specify an end date. The GAMS elements of this set must be in date format.
• Title
  A title is shown when you hover over an item. The GAMS elements of this set can be of any format.
• Group
  When a group variable is specified, elements of the same group are grouped together. The GAMS elements of this set can be of any format.

As an example we use a GAMS Parameter gantt with index sets id, start, end, content and group:

Set
   id       'gannt_id'      / 1, 2, 3, 4, 5, 6 /
   start    'gannt_start'   / '2016-01-04 10:00:00', '2016-01-05 10:30:00', '2016-01-06 10:50:00', '2016-01-07 14:00:00', '2016-01-08 14:45:00', '2016-01-09 19:00:00' /
   end      'gannt_end'     / '2016-01-05 10:00:00', '2016-01-06 10:30:00', '2016-01-07 10:50:00', '2016-01-08 14:00:00', '2016-01-09 14:45:00', '2016-01-10 19:00:00' /
   content  'gannt_content' / test1, test2, test3, test4, test5, test6 /
   group    'gantt_group'   / a, b /
   ;

$onExternalOutput
Parameter gantt(id, start, end, content, group) 'Gantt chart parameter' ;
$offExternalOutput

gantt('1', '2016-01-04 10:00:00', '2016-01-05 10:00:00', 'test1', 'a') = 1;
gantt('2', '2016-01-05 10:30:00', '2016-01-06 10:30:00', 'test2', 'a') = 1;
gantt('3', '2016-01-06 10:50:00', '2016-01-07 10:50:00', 'test3', 'a') = 1;
gantt('4', '2016-01-07 14:00:00', '2016-01-08 14:00:00', 'test4', 'b') = 1;
gantt('5', '2016-01-08 14:45:00', '2016-01-09 14:45:00', 'test5', 'b') = 1;
gantt('6', '2016-01-09 19:00:00', '2016-01-10 19:00:00', 'test6', 'b') = 1;

In a resulting GDX this parameter looks as follows:

In the Configuration Mode we can then configure a Gantt chart.

More information on the tool we use for Gantt charts can be found here.

JSON example

{
  "dataRendering": {
    "gantt": {
      "outType": "graph",
      "graph": {
        "title": "",
        "tool": "timevis",
        "showZoom": true,
        "fit": true,
        "zoomFactor": 0.5,
        "series": {
          "1": {
            "content": "content",
            "start": "start",
            "type": "range",
            "end": "end",
            "title": "content",
            "group": "group",
            "groupTitle": "group"
          }
        },
        "editable": false,
        "multiselect": false,
        "showCurrentTime": false
      },
      "height": 700
    }
  }
}
                                    

Configuration

Example: Model Pickstock, scalars error_train, error_test and error_ratio, singleton sets firstDayTraining and lastDayTraining

Value boxes are special in the sense that only scalar output values (including singleton sets declared as output data) can be displayed as such. The other way around, scalar output values can only be visualized as value boxes in addition to the classic table.

The three scalars to plot represent the absolute error in the entire training phase, the absolute error in the entire testing phase and the ratio between both values.
The scalar error_train is equivalent to the objective function value obj.l. The absolute error in entire testing phase error_test is the sum of all errors in this time: sum(ntd, error(ntd)). The ratio of both values is calculated with error_ratio = error_test/error_train;.
In addition to these numerical values, two singleton sets are used to indicate the dates of the beginning and end of the training period.

$onExternalOutput
Scalar error_train                     'Absolute error in entire training phase'
       error_test                      'Absolute error in entire testing phase'
       error_ratio                     'Ratio between error test and error train';

Singleton Set
       firstDayTraining(date)          'first date of training period'
       lastDayTraining(date)           'last date of training period';
$offExternalOutput

error_train                            = obj.l;
error_test                             = sum(ntd, error(ntd));
if(error_train > 0,
   error_ratio = error_test/error_train;
else
   error_ratio = inf;);

lastDayTraining(td)                    = td.pos = card(td);
firstDayTraining(td)                   = td.pos = 1;

In the Configuration Mode we can configure the value boxes by selecting the symbol Output Scalars:

For a value box, the descriptive text (default: explanatory text of the symbol), the color of the box, an optional icon and the rounding behavior can be configured. The width of a value box results from the number of boxes in a row. Each box can be placed in the desired position by drag & drop. Free rows are ignored.

JSON example

{
  "dataRendering": {
    "_scalars_out": {
      "outType": "valueBox",
      "options": [
        {
          "error_train": {
            "description": "Absolute error in entire training phase",
            "color": "green",
            "icon": {
              "name": "_",
              "lib": "font-awesome"
            },
            "round": 4
          },
          "error_test": {
            "description": "Absolute error in entire testing phase",
            "color": "yellow",
            "icon": {
              "name": "_",
              "lib": "font-awesome"
            },
            "round": 4
          }
        },
        {
          "error_ratio": {
            "description": "Ratio between error test and error train",
            "color": "red",
            "icon": {
              "name": "_",
              "lib": "font-awesome"
            },
            "round": 1
          }
        },
        {
          "firstdaytraining": {
            "description": "first date of training period",
            "color": "aqua",
            "icon": {
              "name": "circle-play",
              "lib": "font-awesome"
            },
            "round": 0
          },
          "lastdaytraining": {
            "description": "last date of training period",
            "color": "aqua",
            "icon": {
              "name": "ban",
              "lib": "font-awesome"
            },
            "round": 0
          }
        }
      ]
    }
  }
}
                                    

A dashboard is an interactive tool with different data views. It is navigated via the ‘tiles’ (value boxes) on the right-hand side of the screen. A data view is shown if the corresponding value box is clicked. It can display multiple charts and tables. Further possible interactions include:

• Interactive filtering of the displayed data
• Toggling between a range of chart/table types
• Exporting the chart/data as a png/csv file

In contrast to other types of visualization, data from several symbols can be displayed in the dashboard (similar to a custom renderer).

Configuration

The dashboard configuration takes place in the dataRendering section of the <modelname>.json file. The configuration includes:

• Configuration of the tiles (value boxes), which are used to navigate between views in the dashboard. The boxes can display scalar values (optional).
• Configuration of the view to be displayed when clicking on a value box. A view can contain several charts/tables.
• Configuration of the individual charts/tables.

The layout of the configuration is as follows:

{
  "dataRendering": {
    "<lowercase_symbolname>": {
      "outType": "dashboard",
      "additionalData": [],
      "options": {
        "valueBoxesTitle": "",
        "valueBoxes": {
            [...]
        },
        "dataViews": {
            [...]
        },
        "dataViewsConfig": {
            [...]
        }
      }
    }
  }
}

The JSON schema, against which the configuration of the "options" object is validated when the application is started, can be found here (search for dashboardOptions in section definitions)

In the following, the dashboard configuration is demonstrated using a parameter macrolevel(mac,sim_miro).

additionalData

A dashboard can display the data of other GAMS input and output symbols that are included in MIRO. For the dashboard to access this data, the corresponding symbols must be specified under "additionalData". Note that only multidimensional symbols should be listed here, as all scalar data is available to the renderer anyway. Example:

{
  "dataRendering": {
    "macrolevel": {
      "outType": "dashboard",
      "additionalData": ["macroperc", "npergdpcomp", "rpergdpcomp", "nperabsorpcomp", "rperabsorpcomp", "qxpc"],
      "options": {
        "valueBoxesTitle": "",
        "valueBoxes": {
            [...]
        },
        "dataViews": {
            [...]
        },
        "dataViewsConfig": {
            [...]
        }
      }
    }
  }
}

valueBoxesTitle

Title to be displayed above the value boxes section in the dashboard.

Example:

{
  "dataRendering": {
    "macrolevel": {
      "outType": "dashboard",
      "additionalData": ["macroperc", "npergdpcomp", "rpergdpcomp", "nperabsorpcomp", "rperabsorpcomp", "qxpc"],
      "options": {
        "valueBoxesTitle": "Summary indicators",
        "valueBoxes": {
            [...]
        },
        "dataViews": {
            [...]
        },
        "dataViewsConfig": {
            [...]
        }
      }
    }
  }
}

valueBoxes

Configuration for the value boxes (tiles) in the dashboard. A value box can display a title and a scalar value (optional). A value box can be linked to a data view that is displayed when the box is clicked.

A value box has several properties that can be configured. Each property of the "valueBoxes" configuration must always be specified so that all arrays have the same length. An example will be given below.

• "id": Each value box needs a unique id. The id of a data view (with charts and tables) is linked to the id of a value box. This can be any string without spaces. Value boxes can also be used without a corresponding view.
• "title": Title that is displayed in the value box.
• "color": Each value box can have an individual color. The color code can either be provided as hexadecimal color code (e.g. '#29B675') or can be chosen from the following options: 'red', 'yellow', 'aqua', 'blue', 'light-blue', 'green', 'navy', 'teal', 'olive', 'lime', 'orange', 'fuchsia', 'purple', 'maroon', 'black'.
• "icon": Each value box can have an individual icon. For all available icons see fontawesome.com.
• "valueScalar": If the value box should display a value, this needs to be configured here. Valid valueScalar entries are scalar output symbols of the model, which are used in MIRO. Names must be specified in lower case.
• "prefix": Any string that is displayed before the value. If '+' is defined as a prefix, a '+' is appended to a positive value. If the value is negative, the sign automatically changes to '-'.
• "postfix": String that is displayed after the value.
• "redPositive": Positive values are displayed in green by default, negative values in red. If this logic should be reversed (positive: red, negative: green), the redPositive property of the value box must be set to true, otherwise false.
• "noColor": If the value of the value box should not have a red/green color, but should be gray instead, the noColor property needs to be set to true, otherwise false.
• "decimals": Number of decimal places that should be displayed for a value.

Example:

{
  "dataRendering": {
    "macrolevel": {
      "outType": "dashboard",
      "additionalData": ["macroperc", "npergdpcomp", "rpergdpcomp", "nperabsorpcomp", "rperabsorpcomp", "qxpc"],
      "options": {
        "valueBoxesTitle": "Summary indicators",
        "valueBoxes": {
          "id": ["summary","gdp","absorption","production"],
          "color": ["#848991","olive","aqua","#3c8dbc"],
          "icon": ["chart-line","money-bill-trend-up","credit-card","industry"],
          "title": ["MACROECONOMIC INDICATORS","GDP","ABSORPTION","PRODUCTION"],
          "prefix": ["+","+","+","+"],
          "postfix": ["%","%","%","%"],
          "redPositive": [false,false,false,false],
          "noColor": [false,false,false,false],
          "valueScalar": [null,"gdp","absorption","production"],
          "decimals": [2,2,2,2]
        },
        "dataViews": {
            [...]
        },
        "dataViewsConfig": {
            [...]
        }
      }
    }
  }
}

dataViews

A data view is shown if the corresponding value box is clicked in the dashboard. It can display multiple charts and tables. In "dataViews" you define how many charts/tables a view should contain. The actual configuration of the charts/tables is done in "dataViewsConfig" (see below).

The id of a data view (e.g. "summary" in the example below) must match the id of a value box in "valueBoxes". This represents the link between the value box and the view. The elements within the ID array (or object) identify the individual charts/tables. After such an id of an individual chart, (e.g. "macro1" in the example below), the title is specified, which is displayed above the chart/table in the dashboard. One can also use the empty string "" to display no header. Note that IDs must not contain spaces.

Example:

{
  "dataRendering": {
    "macrolevel": {
      "outType": "dashboard",
      "additionalData": ["macroperc", "npergdpcomp", "rpergdpcomp", "nperabsorpcomp", "rperabsorpcomp", "qxpc"],
      "options": {
        "valueBoxesTitle": "Summary indicators",
        "valueBoxes": {
          "id": ["summary","gdp","absorption","production"],
          [...]
        },
        "dataViews": {
          "summary": [
            {"macro1": "Macroeconomic indicators (real, percentage change)"},
            {"macro2": ""}
          ],
          "gdp": [
              {"gdp1": "Composition of GDP (nominal, percentage change)"},
              {"gdp2": "Composition of GDP (real, percentage change)"}
          ],
          "absorption": [...],
          "production": [...]
        },
        "dataViewsConfig": {
            [...]
        }
      }
    }
  }
}

dataViewsConfig

All individual charts/tables are configured here. A separate configuration in "dataViewsConfig" is required for each chart listed in the "dataViews". The IDs need to be the same.

A data view configuration describes how the underlying data should be filtered, pivoted, aggregated and which renderer should be used per default. The configuration of a chart is almost the same as that of a configured view in the pivot table (learn more about pivot table views here). The easiest way to configure a dashboard chart is therefore as follows:

1. Start the MIRO app, load a scenario with results data (or solve first).
2. Navigate to the symbol that should be used for the desired chart/table. The renderer of the symbol must be a pivot table (default).
3. In the pivot table, slice and dice data as desired, select a charting type and save the view by clicking on the + on the left side of the pivot table.
   
4. Click on Scenario → Edit metadata in the top right corner of the app and switch to the View tab in the dialog box that opens.
5. Select the configured view (or several views) by clicking on it and click on Download views.
6. Open the downloaded json file and auto-format the contents if necessary. The result can look like this:

   
   {
     "macroperc": {
         "macro1": {
             "aggregationFunction": "sum",
             "pivotRenderer": "bar",
             "domainFilter": { "default": null },
             "tableSummarySettings": {
               "enabled": false, "rowSummaryFunction": "sum", "colSummaryFunction": "sum"
             },
             "rows": "mac",
             "filter": { "sim_miro": "scenario" },
             "chartOptions": { "title": "Macroeconomic indicators (real, percentage change)" }
         }
     }
   }
   

7. Many settings of a view can be copied 1:1 for the configuration of a chart in the dashboard (see example below). Note that the "title" in "chartOptions" is ignored. Instead, the title provided in "dataViews" is used. "domainFilter" is ignored as well.
8. In addition to the attributes of the downloaded pivot-table view, there are further attributes available in the dashboard configuration:
   1. "data": Symbol name that the data should come from. If not provided, it is assumed that the symbol that the renderer is based on should be used.
   2. "userFilter": Adds multi drop down menu(s) to a chart/table in which elements of the configured dimension can be selected for being rendered.
      
      Instead of providing symbol dimensions, one can also set a "dataViewsConfig" ID. As a result, the filter(s) set in the corresponding chart of that ID is/are also applied here. Example:

      {
        "macroperc": {
            "macro1": {
                [...],
                "data": "macroperc",
                "userFilter": "mac"
            },
            "macro2": {
              [...],
              "data": "macroperc",
              "userFilter": "macro1"
          }
        }
      }
      

      In this example, the "macro1" chart will have a filter drop down menu based on all elements in the "mac" dimension of the symbol "macroperc". Filters selected in that chart will automatically be applied to "macro2" as well since it uses "userFilter": "macro1".
   3. "customLabels": Listing of individual elements that should be relabeled in the chart/table. Note that applying custom labels to a dataset can be compute-intensive depending on the amount of data. The time it takes to render the output might increase significantly. It is always a good idea to use appropriate labels in the GAMS model already. Example:

      
      {
        "macroperc": {
            "macro1": {
                [...],
                "chartOptions": {
                  "customLabels": {
                     "a_Agr": "Agriculture",
                     "a_Const": "Construction"
                   }
                }
            }
        }
      }
      

   4. "decimals": Number of decimal places to be displayed in a chart/table.
   5. "colWidth": Specifies the width of the chart. Must be between 1 and 12. If not provided, the chart uses the entire available width (12). If a view contains 2 graphics with a colWidth of 6 each, they are displayed next to each other.
   6. "height": Specifies the height of the chart. Must be any valid css height, e.g. "40vh" or "450px". If not provided, the chart uses 33% of the height of the viewport ("33vh").

Example:

{
  "dataRendering": {
    "macrolevel": {
      "outType": "dashboard",
      "additionalData": ["macroperc", "npergdpcomp", "rpergdpcomp", "nperabsorpcomp", "rperabsorpcomp", "qxpc"],
      "options": {
        "valueBoxesTitle": "Summary indicators",
        "valueBoxes": {
          "id": ["summary","gdp","absorption","production"],
          [...]
        },
        "dataViews": {
          "summary": [
              {"macro1": "Macroeconomic indicators (real, percentage change)"},
              {"macro2": ""}
          ],
          [...]
        },
        "dataViewsConfig": {
          "macro1": {
            "aggregationFunction": "sum",
            "pivotRenderer": "bar",
            "tableSummarySettings": { "enabled": false, "rowSummaryFunction": "sum", "colSummaryFunction": "sum" },
            "rows": "mac",
            "aggregations": { "sim_miro": null },
            "chartOptions": { "title": "Macroeconomic indicators (real, percentage change)" },
            "data": "macroperc",
            "userFilter": "mac",
            "decimals": 2
          },
          "macro2": {
            "aggregationFunction": "sum",
            "pivotRenderer": "table",
            "tableSummarySettings": { "enabled": false, "rowSummaryFunction": "sum", "colSummaryFunction": "sum" },
            "rows": "mac",
            "aggregations": { "sim_miro": null },
            "data": "macroperc",
            "userFilter": "macro1",
            "decimals": 2
          }
        }
      }
    }
  }
}

Besides a chart configuration based on a pivot table view in "dataViewsConfig", one can also provide custom R code, which is rendered at the corresponding position. The dashboard renderer has been prepared in such a way that this requires minimal effort.

In order to add custom code to the dashboard, the entire dashboard renderer must be used as a custom renderer. The following steps are necessary for this:

1. Download the latest dashboard renderer file from the MIRO repository on GitHub and put it into your renderer_<modelname> directory. If the folder does not already exist, you can create it manually.
2. In the dashboard.R file, make the following changes:

   - dashboardOutput <- function(id, height = NULL, options = NULL, path = NULL) {
   + mirorenderer_macrolevelOutput <- function(id, height = NULL, options = NULL, path = NULL) {
       ns <- NS(id)
       [...]
     }
   - renderDashboard <- function(id, data, options = NULL, path = NULL, rendererEnv = NULL, views = NULL, outputScalarsFull = NULL, ...) {
   + renderMirorenderer_macrolevel <- function(input, output, session, data, options = NULL, path = NULL, rendererEnv = NULL, views = NULL, attachments = NULL, outputScalarsFull = NULL, ...) {
   -   moduleServer(
   -     id,
   -     function(input, output, session) {
           ns <- session$ns
           [...]
   
   # These are the last three lines of code in the file
   - }
   -)
   }
    

   In this example the dashboard is rendered for the symbol macrolevel. Make sure to replace macrolevel with your symbol name.
3. In the dataRendering section of the <modelname>.json file change the "outType" of the symbol to render from "dashboard" to "mirorenderer_<symbolname>"

     {
      "dataRendering": {
        "macrolevel": {
   -      "outType": "dashboard",
   +      "outType": "mirorenderer_macrolevel",
          "additionalData": [...],
          "options": {...}
        }
      }
    }
                             

4. That's it. You can now extend the dashboard renderer as a custom renderer. Of course, all the code can be customized here. The aforementioned preparation of the dashboard renderer with regard to custom code instead of a pivot table view can be addressed as follows.

To use custom R code instead of a pivot table view, the "dataViewsConfig" ID in the <modelname>.json file must not be assigned a view (list) as before, but any string, e.g. "macro2": "customCode":

{
  [...]
  "dataViews": {
    "summary": [
        {"macro1": "Macroeconomic indicators (real, percentage change)"},
        {"macro2": ""}
    ]
  },
  "dataViewsConfig": {
    "macro1": {...},
    "macro2": "customCode"
  }
}

Note that the title that has been provided in the corresponding "dataViews" configuration is ignored in this case.

Custom R code can now be written in the renderer file, which is then rendered in the dedicated data view. The ID of the chart is used as the ID for the output function. Make sure that you place the custom code within the renderer function (here: within renderMirorenderer_macrolevel{}). Example:

renderMirorenderer_macrolevel <- function(input, output, session, data, options = NULL, path = NULL, rendererEnv = NULL, views = NULL, attachments = NULL, outputScalarsFull = NULL, ...) {
  [...]

  output[["macro2"]] <- renderUI({
    tagList(
      column(12,
             tags$h1("This is custom code",
                     style ="height: 33vh;text-align: center;background-color: lightblue;line-height: 33vh;")
      )
    )
  })
}

The output is as shown in the screenshot above. Of course, the entire view can also be customized instead of just part of it:

{
  [...]
  "dataViews": {
    "summary": [
        {"macro2": ""}
    ]
  },
  "dataViewsConfig": {
    "macro2": "customCode"
  }
}
                        

Tip:

If you are using the dashboard with custom code, it is best to take a look at the chapter on custom renderers too.

Configuration

Section Custom graphics gives examples and explains in detail how to deal with custom renderers in general.

The Configuration Mode can help you create your own renderers. If a symbol is selected for the use of a custom renderer in Configuration Mode, the following screen appears:

With the custom renderer editor you can write your own renderer and see the result immediately. The upper editor defines the placeholder or output function, while the lower editor is for the renderer function. If you are not yet familiar with the concept of output and render function, please read the corresponding section first. When you click the "Update" button, the code is executed in both cells and the graph appears on the right (or an error message if you made a mistake). As an example consider the map created for the transport model below

To save your renderer, click the "Save" button at the bottom left. If it does not already exist, MIRO creates the folder renderer_<modelname> in your current working directory and saves your code in a file called: mirorenderer_<symbolname>.R.

On the Advanced Options tab, you can add additional symbols to be used by your renderer, specify R packages that are not included in MIRO but are required by your renderer, and set additional options that will be available in your renderer (read here to learn more). Additional options are especially useful when you share your renderer between different symbols.

The resulting JSON snippet for our transport example looks like this:

{
   "dataRendering":{
      "schedule": {
         "outType": "transportMap",
         "packages": "leaflet",
         "additionalData": ["ilocdata", "jlocdata"],
         "options":{
            "title":"Optimal transportation schedule"
         }
      }
   }
}
                                    

As already mentioned, not all the options available for the different graphs are explained here. Nevertheless, some options are given as examples, especially those that are applicable to several graphic types.

Group domain

With some graphic types, a symbol dimension can be internally pivoted. This makes the graphic independent of the data.

In the model Pickstock there is the parameter Price, which contains all stock symbols of the Dow Jones index with their daily prices for a period of one year:

Set       date                 'date'
        symbol               'stock symbol';
$onExternalInput
Parameter price(date<,symbol<) 'Price';
$offExternalInput
 

Price is not declared as a table but as a normal parameter. For MIRO, this means that the data is available in list format with the headers date and symbol. The individual stocks are listed in the rows:

We want to configure a line chart in which the price trend of each stock symbol is plotted separately. Now MIRO needs the price information of each symbol for each date. The problem is that MIRO only knows about one value column: Price. Hence, we cannot distinguish between the prices of individual stocks. In the resulting line chart a single line is plotted which contains all prices of all stock symbols:

In the section Filters & Domains we can can select a domain we want to pivot internally. If we select the domain stock symbol here, the tool plots a line for each individual stock symbol:

With this option, the chart configuration is completely independent of the data. We could even import data from another stock index - say the French CAC40 - and plot these accordingly.

The internal pivoting is available for the chart types Bar chart, Scatter plot, Line chart, Bubble chart, Histogram and Time series diagram.

Please also see the chapter about tables in MIRO to understand the difference between the MIRO-internal pivoting done here and the GAMS Table statement.

Filter graph

One dimension of a symbol visualized as a graphic can be configured as a filter dimension. Its elements can then be selected via a (multi-)drop-down menu. Only the data belonging to this selection is displayed in the graphic.

Example: Model Pickstock, parameter price

The Parameter price has the dimensions date and symbol: Parameter price(date,symbol) 'Price';. With the previous option Group domain we made sure that each stock is displayed as a separate line (and the graphic is therefore independent of the data). Now we can activate the checkbox Should one domain of the symbol be selectable via a dropdown menu? in the same category Filters & Domains and select the dimension stocks as filter dimension. In the result a (multi-) dropdown menu appears above the chart, in which we can select the individual stocks. Only those stocks are displayed which are selected in the filter.

Tip:

If your filter dimension contains data in date format (e.g. 2020-05-06), you will see a date range selector instead of a multi-dropdown menu for selection.

Show graphic and data table in split screen

If you have configured a graphical representation of your data, you can always switch between plot and data table by clicking on the button in the upper right corner:

With the option Show graphic and data table in split screen in the section General in Configuration Mode, both data representations can be displayed next to each other:

Animation

Chart data can be animated in order to show how the data changes over time. Example: We want to animate a chart that shows the individual stocks of the Dow Jones index over time. The Y axis shows the price of a stock, the X axis shows the date. Now we want to display all stocks one after the other with an animation. For this we select in the category Y axix the dimension, which should be used in the animation for the single frames. Since we want to show all stock symbols one after the other, we choose stock symbol:

After the selection the animation is rendered and a slider with play button appears under the graphic.
Note: The animation of graphics can require a lot of computing power. It is therefore possible that it takes a little time until the preview graphics are rendered. This is especially true if you select a dimension for animation that consists of many individual values.

Under Animation we can further customize the animation, for example by setting the number of frames per second or the type of transition between two frame transitions.

Tip:

When using animations, it is often useful to define the visible areas for both axes in the corresponding axis categories. Otherwise it can happen that areas are pre-selected in which no data is visible.