0

HTML Pivot Table

Pivot Table.ninox

Hello everyone,

In my application, time series data (screenshot 1) are often needed to be managed and presented in a practical way.

Since Ninox does not yet allow us to insert a pivot table view into a page, I thought of creating an HTML table for time series data (screenshot 2).

Below is the code inside a function field:

let x := Covid19_Data;
let y := rsort(unique(x.date));
let header := for loop1 in y do
        "<th>" + loop1 + "</th>"
    end;
let body := for loop2 in unique(x.country) do
        "<tr>
        <td style='background-color: #ffffff'>" +
        loop2 +
        "</td>" +
        for loop3 in y do
            let z := last(x[country = loop2 and date = loop3]);
            "<td><span onclick=""ui.popupRecord('" + raw(z.Id) +
            "')""; style=""cursor: pointer;"">" +
            z.daily_new_cases +
            "</span></td>"
        end +
        "</tr>"
    end;
html("<head>
<style>
    .pivot-table {
        border-collapse: collapse;
        width: auto;
        table-layout: auto; /* Column width based on content */
        white-space: nowrap; /* Prevent text wrapping */
    }
    .pivot-table th, .pivot-table td {
        border: 1px solid #dddddd;
        text-align: center;
        padding: 8px;
    }
    .pivot-table th {
        background-color: #ffffff;
        position: sticky;
        top: 0;
        z-index: 2; /* Higher z-index to ensure it's above other elements */
    }
    .pivot-table th:first-child {
        background-color: #ffffff;
        font-weight: bold;
        position: sticky;
        left: 0;
        z-index: 3; /* Higher z-index to ensure it's always on top */
        width: 200px; /* Custom width for the first column */
    }
    .pivot-table td:first-child {
        background-color: #ffffff;
        font-weight: bold;
        position: sticky;
        left: 0;
        z-index: 1; /* Lower z-index to ensure it's below the first column header */
        max-width: 200px; /* Set maximum width for the first column */
        overflow: hidden; /* Hide overflow content */
    }
</style>
</head>
<body>
<div>
<table id='pivotTable' class='pivot-table'>
    <tr>
        <th></th>" +
header +
"
    </tr>" +
body +
"
</table>
</div>
</body>")

 

Although the above code works very well with small datasets, it has performance issues with large ones.

For example, the attached sample database (9944 records) takes about 20 seconds to fetch the data.

Any suggestions for improvement from the experts are welcome 🙂.

20 replies

null
    • szormpas
    • 7 mths ago
    • Reported - view

    UPDATE: 10X speed improvement!

    I improved the speed of data display by at least ten times, from about 20 seconds to less than 2 seconds, by removing the nested  “for” loop inside the body variable 😀.

    let x := (Covid19_Data order by -date);
let header := for loop1 in rsort(unique(x.date)) do
        "<th>" + loop1 + "</th>"
    end;
let body := for loop2 in unique(x.country) do
        let y := x[country = loop2];
        "<tr>
        <td style='background-color: #ffffff'>" +
        loop2 +
        "</td>" +
        y.("<td><span onclick=""ui.popupRecord('" + raw(Id) +
        "')""; style=""cursor: pointer;"">" +
        daily_new_cases +
        "</span></td>") +
        "</tr>"
    end;
html("<head>
<style>
    .pivot-table {
        border-collapse: collapse;
        width: auto;
        table-layout: auto; /* Column width based on content */
        white-space: nowrap; /* Prevent text wrapping */
    }
    .pivot-table th, .pivot-table td {
        border: 1px solid #dddddd;
        text-align: center;
        padding: 8px;
    }
    .pivot-table th {
        background-color: #ffffff;
        position: sticky;
        top: 0;
        z-index: 2; /* Higher z-index to ensure it's above other elements */
    }
    .pivot-table th:first-child {
        background-color: #ffffff;
        font-weight: bold;
        position: sticky;
        left: 0;
        z-index: 3; /* Higher z-index to ensure it's always on top */
        width: 200px; /* Custom width for the first column */
    }
    .pivot-table td:first-child {
        background-color: #ffffff;
        font-weight: bold;
        position: sticky;
        left: 0;
        z-index: 1; /* Lower z-index to ensure it's below the first column header */
        max-width: 200px; /* Set maximum width for the first column */
        overflow: hidden; /* Hide overflow content */
    }
</style>
</head>
<body>
<div>
<table id='pivotTable' class='pivot-table'>
    <tr>
        <th></th>" +
header +
"
    </tr>" +
body +
"
</table>
</div>
</body>")
      • Fred
      • 7 mths ago
      • Reported - view

      , great news!

      I was testing out on my table of 11,000+ records and I was getting about 5 seconds even with a select. I didn't have any loops, but I did create my own JSON table that I used to create the table rows instead of a for loop.

      Something like:

      let x := (select GAMESTATS);
let data := x.{
        dateOfGame: BOXSCORES.Date,
        city: BOXSCORES.PARK.CITY,
        team: Teams.Name,
        runs: RUNS,
        hits: HITS,
        bAvg: round(HITS / ATBATS, 1)
    };
let thead := some code;
let tbody := data.("...rest of code

      The new JSON table allowed me to use do calculations and such (see bAvg). I'm guessing it won't help too much as you already have it down to 5 seconds for 9,000+ records.

    • szormpas
    • 7 mths ago
    • Reported - view

    Thank you , I appreciate your insights.

    While I understand the benefits of creating a new JSON object, in my specific case, the data is already there and requires no further configuration.

    So far, I have managed to get rid of the second loop, so I am down to one. How could I have created the table data without at least one loop?

    With the changes to the code, it takes about 30 seconds to display 184,787 records so far.

    The new database is attached for anyone interested in experimenting with this extensive dataset.

    let x := ((Covid19_Data order by -date) order by country);
let tdata := "<tr><th></th>" + unique(x.("<th>" + date + "</th>")) + "</tr>" +
    for loop1 in unique(x.country) do
        let y := x[country = loop1];
        "<tr>
            <td style='background-color: #ffffff'>" +
        loop1 +
        "</td>" +
        y.("<td><span onclick=""ui.popupRecord('" + raw(Id) +
        "')""; style=""cursor: pointer;"">" +
        daily_new_cases +
        "</span></td>") +
        "</tr>"
    end;
html("<head>
<style>
    .pivot-table {
        border-collapse: collapse;
        width: auto;
        table-layout: auto; /* Column width based on content */
        white-space: nowrap; /* Prevent text wrapping */
    }
    .pivot-table th, .pivot-table td {
        border: 1px solid #dddddd;
        text-align: center;
        padding: 8px;
    }
    .pivot-table th {
        background-color: #ffffff;
        position: sticky;
        top: 0;
        z-index: 2; /* Higher z-index to ensure it's above other elements */
    }
    .pivot-table th:first-child {
        background-color: #ffffff;
        font-weight: bold;
        position: sticky;
        left: 0;
        z-index: 3; /* Higher z-index to ensure it's always on top */
        width: 200px; /* Custom width for the first column */
    }
    .pivot-table td:first-child {
        background-color: #ffffff;
        font-weight: bold;
        position: sticky;
        left: 0;
        z-index: 1; /* Lower z-index to ensure it's below the first column header */
        max-width: 200px; /* Set maximum width for the first column */
        overflow: hidden; /* Hide overflow content */
    }
</style>
</head>
<body>
<div>
<table id='pivotTable' class='pivot-table'>
" +
tdata +
"
</table>
</div>
</body>")
    Pivot_Table_2.ninox
    • Fred
    • 7 mths ago
    • Reported - view
     said:
    it takes about 30 seconds to display 184,787 records so far.

     whaaaat? 184K records?

    wow

    it took 9 secs to load on my M2 MBPro.

      • szormpas
      • 7 mths ago
      • Reported - view

         9 secs...so far so good!

      • Fred
      • 7 mths ago
      • Reported - view

      I didn't do anything. I just loaded the table and it took 9 seconds on my computer in the MacOS app. I don't have a cloud account that can handle that many records. :(

      • Rafael Sanchis
      • Rafael_Sanchis
      • 7 mths ago
      • Reported - view

      On Clound take a long time, more than 35 seconds. ( Surface Pro 7)

      • szormpas
      • 7 mths ago
      • Reported - view

        I actually use the public cloud and the secs I mention are for that. What initially attracted me to Ninox was the idea of a centralised database in the cloud, so I expect it to be able to handle such large datasets with relative ease!

      • szormpas
      • 7 mths ago
      • Reported - view

        It's possible that the difference in seconds is also due to the speed of everyone's internet connection.

      • Fred
      • 7 mths ago
      • Reported - view

      my test was done local on my machine. These MBP are fast. 🙂

      • Rafael Sanchis
      • Rafael_Sanchis
      • 7 mths ago
      • Reported - view

       

      Maybe my speed is 650 to 700 Mbps

    • szormpas
    • 7 mths ago
    • Reported - view

    I have tried tweaking the code further by using "do as server," "do as transaction," and "cached()," but nothing seems to improve things. 

    Maybe it's time to consider "pagination." This is an entirely new area for me, but you can only build an HTML table if you consider its performance when using large datasets 😥.

      • Fred
      • 7 mths ago
      • Reported - view

      I was wondering if do as server would help, but then when building html I don't know if it will show anything. but it seems like it didn't make much of a difference.

      • szormpas
      • 7 mths ago
      • Reported - view

        you are right, I did use the "do as server" and  the html() showed nothing!

    • John_Halls
    • 7 mths ago
    • Reported - view

    My computer at work struggles to display this data. I get an 'out of memory' message

    • Rafael Sanchis
    • Rafael_Sanchis
    • 7 mths ago
    • Reported - view

    I use this code

    let me := this;
let Rows := unique(('History Hours' order by 'Staff:').'Staff:');
let Cols := unique(('History Hours' order by yw).yw);
"--------------CSS---------------";
let css := "<style>
table,th, td {
    border:thin solid #a0a0a0;
    border-collapse:collapse;
}
th{
    background-color:#f1f3f4;
    text-align:center;
    padding:0.1em  0.6em;
}
tr:nth-child(odd) {background-color:#e4ebf2;}
td{
    text-align:right;
    padding:0.1em  0.8em;
    font-family:Courier;
}  th:nth-of-type(1) {
    position:sticky;
    overflow-x:auto;
    white-space:nowrap;
    left:-1em;
    text-align:left;
}
.grey {background-color:silver;}
</style>";
"------------Content------------";
let content := "
<table>
<tr>
    <th class ='stop'>Year / Week</th>
" +
    for c in Cols do
        "<td>" + c + "</td>"
    end +
    "<td>Total</td>
</tr>
" +
    for r in range(0, length(Rows)) do
        "<tr><th>" + item(Rows, r) + "</th>" +
        for c in range(0, length(Cols)) do
            "<td>" +
            sum(me.'History Hours'['Staff:' = item(Rows, r) and yw = item(Cols, c)].'Hours:') +
            "</td>"
        end +
        "<td class = 'grey'>" +
        sum(me.'History Hours'['Staff:' = item(Rows, r)].'Hours:') +
        "</td>
        </tr>"
    end +
    "
<tr><th>Total</th>" +
    for c in Cols do
        "<td class = 'grey'>" + sum(me.'History Hours'[yw = c].'Hours:') + "</td>"
    end +
    "<td class = 'grey'>" +
    sum(me.'History Hours'.'Hours:') +
    "</td></tr>
</table>";
html(css + content)
    • Rafael Sanchis
    • Rafael_Sanchis
    • 7 mths ago
    • Reported - view

    Follow this 

    https://forum.ninox.de/t/83h4jdf/html-crosstab

    • szormpas
    • 7 mths ago
    • Reported - view

     your table is nice and offers new features for integration (e.g Choice Options). It looks like you are also using an HTML table with the same logic (you can call it Pivot or Crosstab).

    I see in your code that you're also using a nested for loop, so I'm not expecting better load times.

    So the question remains, what is the most efficient way to create a pivot or crosstab table so that it can handle a large amount of data?

    • Fred
    • 7 mths ago
    • Reported - view

    So I've been racking my brain on how to make things go faster and one way I thought of is to create a table called byCountry, then link all of the Covid19 data to their respective country. That way I am using the power of relationships and not trying to sort/filter 184K records.

    Check out the DataPresentation table in the attached DB.

    Here are the load times on my MBP M2 running 3.12.2 app and db stored on iCloud (because I don't have a cloud account that can load 184K records).

     

    Time

     

    1st

    2nd

    3rd

    Dashboard

    6.3

    7.0

    7.2

    Data Presentation

    3.2

    3.4

    2.6
    largeDateSetV1.ninox
      • szormpas
      • 7 mths ago
      • Reported - view

       Hi, this is an interesting approach that I haven't been though about, breaking down the ininial dataset into one or more table if feasible.

      Indeed, I tested DataPresentation table and I can confirm that I noticed a slight improvement of a few seconds on public cloud using MP air M1.

      One thing is certain. The public cloud is much slower than a local database in the Mac app.

      I wonder if the private cloud offers much better speeds.

Login to reply

Content aside

  • 7 mths agoLast active
  • 20Replies
  • 549Views
  • 4 Following