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.

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 comes with a powerful pivot table renderer that allows you to filter, aggregate and pivot data. In addition to a table, the MIRO pivot table renderer also supports the visualization of data in the form of a heatmap or as chart (line chart, bar chart, etc.). Each slice, or "view" of data can be exported by clicking the button. If you are currently viewing a chart, the view is exported as a PNG file; if you are viewing a table or heatmap, the data is exported as a CSV file.

A click on the button toggles the presentation mode.

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. 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.

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 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
          }
        }
      ]
    }
  }
}
                                    

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.