Metrics/KPI and Data tables¶
Charts, Maps, tables, calendar are default views. Lot of additionnal parameters are waiting for you in the documentation. In the mean time, a lot of other widgets are as insterested as the main ones.
Please have a look to the widget list for more information.
Pick up a dataset¶
To illustrate metrics, KPI, advanced map tooltips and more, we will use a dataset that contains all regularity reports from the French Railway Society since 2011. This dataset contains, for each TGV (High speed train) line (100 lines), the number of scheduled train per month, the number of trains that didn’t leave the station, and the number of trains in late.
Dataset id : regularite-mensuelle-tgv
You can search for it on data.opendatasoft.com
Then add this dataset to your domain (New dataset, Add a source from OpenDataSoft network, pick the dataset, save, publish)
Create some KPIs¶
To create metrics / KPI, we use odsAggregation, to compute the sum, average, min, max of a data field. Based on the context, the metric will update dynamically all along the user navigation (queries, filters).
Starting from a new page, let’s start by getting a context from this dataset. The simplest way is to go to the explore/catalog page, go to the dataset, then in the table view, copy and paste de widget code from the share links below :
<ods-dataset-context context="regularitemensuelletgv" regularitemensuelletgv-dataset="regularite-mensuelle-tgv"> <ods-table context="regularitemensuelletgv"></ods-table> </ods-dataset-context>
On top of the table view, we will add a counter based on the nombre_de_trains_programmes field (number of scheduled train).
Let’s get the number of trains scheduled since 2011 in France.
- With odsAggregation, the parameters are :
- the name of the variable that will store aggregation result
- the context to work on
- the function
- the expression (the name of the field)
With a number filter to display it correctly in an HTML span you should have something similar to this :
<div class="container"> <ods-dataset-context context="regularitemensuelletgv" regularitemensuelletgv-dataset="regularite-mensuelle-tgv"> <div ods-aggregation="myvar" ods-aggregation-context="regularitemensuelletgv" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_programmes"> <span>Scheduled train : {{ myvar | number }}</span> </div> <br/> <ods-table context="regularitemensuelletgv"></ods-table> </ods-dataset-context> </div>
It displays a number (approx. 1.7 millions) at the beginning of the page that already need some CSS to be highlighted. Metric are often rounded in a box, with an important font size and weight.
Add this CSS code to your page :
.kpi { padding: 2px 30px; /* Give some air to the digits */ border: 1.5px solid #010201; /* the border */ border-radius: 5px; /* rounded corners */ font-size: 2em; /* bigger font */ font-weight: 500; /* thicker font */ margin: 10px 5px; /* give some air around the KPI */ }
And add kpi class to your span tag.
Save, refresh, and see that it’s a bit better.
Before going further with CSS and specially repsonsive code, let’s add 2 more metrics : number of cancelled train (nombre_de_trains_annules) and number of delayed train (nombre_de_trains_en_retard_a_l_arrivee)
<div ods-aggregation="myvar" ods-aggregation-context="regularitemensuelletgv" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_programmes"> <span class="kpi">Scheduled train : {{ myvar | number }}</span> </div> <br/> <div ods-aggregation="myvar" ods-aggregation-context="regularitemensuelletgv" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_annules"> <span class="kpi">Canceled train : {{ myvar | number }}</span> </div> <br/> <div ods-aggregation="myvar" ods-aggregation-context="regularitemensuelletgv" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_en_retard_a_l_arrivee"> <span class="kpi">Delayed train : {{ myvar | number }}</span> </div> <br/> <ods-table context="regularitemensuelletgv"></ods-table>
Result :
It works but all metrics are independant. We can add filters to make them become dynamic depending on user filters etc... but we can’t compute any thing from it. It would be interesting to have the pourcentage of canceled or delayed train out of the total.
To do this, you must chain ods-aggregation with different variable names.
It must looks like that :
<div ods-aggregation="total" ods-aggregation-context="regularitemensuelletgv" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_programmes"> <div ods-aggregation="canceled" ods-aggregation-context="regularitemensuelletgv" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_annules"> <div ods-aggregation="delayed" ods-aggregation-context="regularitemensuelletgv" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_en_retard_a_l_arrivee"> <span class="kpi">Scheduled : {{ total | number }}</span> <span class="kpi">Canceled : {{ canceled | number }}</span> <span class="kpi">Delayed : {{ delayed | number }}</span> </div> </div> </div>
Then, it’s now possible, with AngularJS expression, to compute pourcentages between each values :
<span class="kpi">Scheduled : {{ total | number }}</span> <span class="kpi">Canceled : {{ canceled | number }}</span> <span class="kpi">Delayed : {{ delayed | number }}</span> <span class="kpi">% Canceled : {{ canceled / total * 100 | number : 2 }}%</span> <span class="kpi">% Delayed : {{ delayed / total * 100 | number : 2 }}%</span>
Note
| number : 2 is the AngularJS filter to pretty print numerical values. The optional parameter 2 is to limit to only 2 decimals
Before going further let’s clean up the HTML code to be able to apply CSS for responsive display, different font size between the title and the value etc... :
<div class="row"> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> Scheduled </div> <div class="kpi-value"> {{ total | number }} </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> Canceled </div> <div class="kpi-value"> {{ canceled | number }} </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> Delayed </div> <div class="kpi-value"> {{ delayed | number }} </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> % Canceled </div> <div class="kpi-value"> {{ canceled / total * 100 | number : 2 }}<span class="kpi-value-unit"> %</span> </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> % Delayed </div> <div class="kpi-value"> {{ delayed / total * 100 | number : 2 }}<span class="kpi-value-unit"> %</span> </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> On time </div> <div class="kpi-value"> {{ total - delayed - canceled | number }} </div> </div> </div> </div>
And replace the CSS :
.kpis { display: inline-flex; } .kpi { text-align: center; padding: 5px 0px; margin-bottom: 10px; height: 70px; border: 1.5px solid #010201; /* the border */ border-radius: 5px; /* rounded corners */ } .kpi-title { font-size: 1em; /* bigger font */ font-weight: 400; /* thicker font */ } .kpi-value { font-size: 2em; /* bigger font */ font-weight: 500; /* thicker font */ } .kpi-value-unit { font-size: 1.5rem; /* bigger font */ font-weight: 400; /* thicker font */ }
Save, refresh, see :
Colorize your KPI - ng-class¶
ng-class directive is very conveniant when an HTML element must be styled depending on a value, the context, an expression or all other dynamic things that can be used in AngularJS expression.
For exemple, to define a threshold on our metrics, green if the pourcentage of canceled trains is below 0,20% and red is it’s bigger, add this to your element :
<div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi" ng-class="{'good' : canceled / total * 100 < 0.2, 'bad' : canceled / total * 100 >= 0.2}"> <div class="kpi-title"> % Canceled </div> <div class="kpi-value"> {{ canceled / total * 100 | number : 2 }}<span class="kpi-value-unit"> %</span> </div> </div> </div>
And this CSS :
.good { color: #55cd61; border-color: #55cd61; } .medium { color: #ff9c22; border-color: #ff9c22; } .bad { color: #e50000; border-color: #e50000; }
Save, refresh, and see that our metrics is now red (the bad CSS class is used)
- To clearly see the dynamic behavior, let’s add some filters on the left of the table (on the second line). To do so :
- add a
divwith therowclass to surround the table - add a
divwithcol-md-9around the table - add a
ods-facetswidgets with 2ods-facetfor filtersdateanddepart - surround
ods-facetswith adivstyled withcol-md-3 - optionnaly add some
ods-boxstyleddivto surround blocks with a white background
It should looks like this :
<div class="row"> <div class="col-md-3"> <div class="ods-box"> <ods-facets context="regularitemensuelletgv"> <h2> Date </h2> <ods-facet name="date"></ods-facet> <h2> Origin station </h2> <ods-facet name="depart"></ods-facet> </ods-facets> </div> </div> <div class="col-md-9"> <div class="ods-box"> <ods-table context="regularitemensuelletgv"></ods-table> </div> </div> </div>
- add a
Save, refresh, and see how it behave when filters are selected, In 2013, Paris EST Station had less than 0,20% of canceled train. Have it a look.
Note
ng-classsyntax is, surround by brackets, the CSS class in quotes followed by:then the AngularJS expression :{ 'CSS class' : AngularJS expression, ... }Several classes and expression can be specified, all the expressions will be tested, from left to right.
Compare with filters : highlight differences with colors¶
It’s interresting to compare the full dataset, and a filtered view by the user and see if the metrics (specially pourcentages) are differents, bigger or smaller.
The plan is to have 2 contexts, one that will represent the full dataset, and one that can be filtered by the user. Once a filter is applied, if the metric is bigger or smaller that the full dataset metric, it will be highlighted by a color.
- To do so, you will need :
- a secondary context
- plug
ods-facetsandods-tableon this secondary context - compute the same metrics for both contexts
- add the secondary metric IF and ONLY IF a filter is selected
- configure the
ng-classto compare full and filtered metric
Let’s go.
Two contexts, one dataset :
<ods-dataset-context context="regularitemensuelletgv,regularitemensuelletgvfiltered" regularitemensuelletgv-dataset="regularite-mensuelle-tgv" regularitemensuelletgvfiltered-dataset="regularite-mensuelle-tgv">
Metrics full blocks should looks like this now :
<div ods-aggregation="total" ods-aggregation-context="regularitemensuelletgv" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_programmes"> <div ods-aggregation="canceled" ods-aggregation-context="regularitemensuelletgv" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_annules"> <div ods-aggregation="delayed" ods-aggregation-context="regularitemensuelletgv" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_en_retard_a_l_arrivee"> <div ods-aggregation="totalfiltered" ods-aggregation-context="regularitemensuelletgvfiltered" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_programmes"> <div ods-aggregation="canceledfiltered" ods-aggregation-context="regularitemensuelletgvfiltered" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_annules"> <div ods-aggregation="delayedfiltered" ods-aggregation-context="regularitemensuelletgvfiltered" ods-aggregation-function="SUM" ods-aggregation-expression="nombre_de_trains_en_retard_a_l_arrivee"> <div class="row"> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> Scheduled </div> <div class="kpi-value"> {{ totalfiltered | number }} </div> <div class="kpi-value-reference"> ({{ total | number }}) </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> Canceled </div> <div class="kpi-value"> {{ canceledfiltered | number }} </div> <div class="kpi-value-reference"> ({{ canceled | number }}) </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> Delayed </div> <div class="kpi-value"> {{ delayedfiltered | number }} </div> <div class="kpi-value-reference"> ({{ delayed | number }}) </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> % Canceled </div> <div class="kpi-value"> {{ canceledfiltered / totalfiltered * 100 | number : 2 }}<span class="kpi-value-unit"> %</span> </div> <div class="kpi-value-reference"> ({{ canceled / total * 100 | number : 2 }}<span class="kpi-value-unit"> %</span>) </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> % Delayed </div> <div class="kpi-value"> {{ delayedfiltered / totalfiltered * 100 | number : 2 }}<span class="kpi-value-unit"> %</span> </div> <div class="kpi-value-reference"> ({{ delayed / total * 100 | number : 2 }}<span class="kpi-value-unit"> %</span>) </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi"> <div class="kpi-title"> On time </div> <div class="kpi-value"> {{ totalfiltered - delayedfiltered - canceledfiltered | number }} </div> <div class="kpi-value-reference"> ({{ total - delayed - canceled | number }}) </div> </div> </div> </div> </div> </div> </div> </div> </div> </div>
Adapt CSS by updating or adding these classes :
.kpi { height: 90px; /* ... */ } .kpi-value-reference { font-size: 1em; /* bigger font */ font-weight: 400; /* thicker font */ } .kpi-value-unit { font-size: 0.7em; /* bigger font */ /* ... */ }
Finally, let’s colorize % delayed and % canceled metrics, if the filtered result is bigger than the total average, show it in red, else, if it’s smaller (better) show it in green.
<div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi" ng-class="{ 'good': (canceledfiltered / totalfiltered * 100) < (canceled / total * 100), 'bad': (canceledfiltered / totalfiltered * 100) > (canceled / total * 100), }"> <div class="kpi-title"> % Canceled </div> <div class="kpi-value"> {{ canceledfiltered / totalfiltered * 100 | number : 2 }}<span class="kpi-value-unit"> %</span> </div> <div class="kpi-value-reference"> ({{ canceled / total * 100 | number : 2 }}<span class="kpi-value-unit"> %</span>) </div> </div> </div> <div class="col-md-2 col-sm-3 col-xs-4"> <div class="kpi" ng-class="{ 'good': (delayedfiltered / totalfiltered * 100) < (delayed / total * 100), 'bad': (delayedfiltered / totalfiltered * 100) > (delayed / total * 100), }"> <div class="kpi-title"> % Delayed </div> <div class="kpi-value"> {{ delayedfiltered / totalfiltered * 100 | number : 2 }}<span class="kpi-value-unit"> %</span> </div> <div class="kpi-value-reference"> ({{ delayed / total * 100 | number : 2 }}<span class="kpi-value-unit"> %</span>) </div> </div> </div>
Save, refresh, and test with year 2015 :
Create a data table¶
odsAnalysis get the analysis of one or several aggregation for each values if a field.
It can be seen exactly like a chart, but instead of displaying curves, columns or pies, odsAnalysis returns each values, and you are free to display them in a table or other expression.
First, we will test the result by seting up the correct analysis in the analysis tab of the dataset. What could be interresting is to display the table of the top 10 regularity stations. To do so :
- X axis : “départ” (origin station)
- 4 series : - AVG Regularity (Régularité) - AVG Number of shceduled train (nombre_de_trains_programmes) - AVG Number of canceled train (nombre_de_trains_annules) - AVG Number of delayed train (nombre_de_trains_en_retard_a_l_arrivee)
- sort by regularity serie
- Number of point : 10 max
The aim is to reproduce this exact analysis with odsAnalysis. As a reminder, please visit the documentation here
odsAnalysisparameters :- ods-analysis : the name of the variable
- ods-analysis-context : the context to work on
- ods-analysis-max : 10 elements max
- ods-analysis-x : the field to work on (all aggregation will be computed for each values of this field)
- ods-analysis-serie-xxx : define a serie called xxx based on an expression (the field) and a function (AVG, MAX, MIN, SUM etc...)
- ods-analysis-sort : sort on a serie by providing the name or -name to invert the sorting
Before jumping into the code (HTML) have a look to the result stored into the variable :
<div class="row"> <div ods-analysis="results" ods-analysis-context="regularitemensuelletgvfiltered" ods-analysis-max="10" ods-analysis-x="depart" ods-analysis-serie-regularity="AVG(regularite)" ods-analysis-serie-scheduled="AVG(nombre_de_trains_programmes)" ods-analysis-serie-canceled="AVG(nombre_de_trains_annules)" ods-analysis-serie-delayed="AVG(nombre_de_trains_en_retard_a_l_arrivee)" ods-analysis-sort="regularity" > {{ results }} </div> </div>
Pretty print the json block :
{ "results":[ { "scheduled":97.37288135593221, "canceled":0.2033898305084746, "regularity":94.60508474576272, "delayed":5.288135593220339, "x":"ST MALO" }, { "scheduled":283.728813559322, "canceled":1.2203389830508475, "regularity":94.28813559322033, "delayed":16.135593220338983, "x":"NANCY" }, { "scheduled":160.64406779661016, "canceled":0.559322033898305, "regularity":92.61186440677963, "delayed":11.677966101694915, "x":"QUIMPER" } ] }
We have a results json list, each block has an x value, the origin station, and 4 aggregations/values that correspond to our series.
AngularJS directive ng-repeat allow to iterate over the results list. For each element, we print a new line in the table.
<div ods-analysis="tgvanalysis" ods-analysis-context="regularitemensuelletgvfiltered" ods-analysis-max="10" ods-analysis-x="depart" ods-analysis-serie-regularity="AVG(regularite)" ods-analysis-serie-scheduled="AVG(nombre_de_trains_programmes)" ods-analysis-serie-canceled="AVG(nombre_de_trains_annules)" ods-analysis-serie-delayed="AVG(nombre_de_trains_en_retard_a_l_arrivee)" ods-analysis-sort="regularity" > <table id="top10"> <thead> <tr> <td>Position</td> <td>Train station</td> <td>Regularity</td> <td>Scheduled</td> <td>Canceled</td> <td>Delayed</td> </tr> </thead> <tbody> <tr ng-repeat="(i, result) in tgvanalysis.results"> <td> {{ i + 1 }} </td> <td> {{ result.x }} </td> <td> {{ result.regularity | number : 2 }} </td> <td> {{ result.scheduled | number : 2 }} </td> <td> {{ result.canceled | number : 2 }} </td> <td> {{ result.delayed | number : 2 }} </td> </tr> </tbody> </table> </div>
Some CSS to render the HTML table nicely :
#top10 { margin: 20px auto; } #top10 thead > tr { background-color: #007396; color: white; } #top10 td { min-width: 100px; padding: 3px 10px; text-align: center; } #top10 tr:nth-child(even) { background-color: #ededed; } #top10 tr:hover{ background-color:#ccc; }
Save, refresh, play with filters :
