A step by step walkthrough for plotting two numeric columns from your own spreadsheet in PlotSet, grouping the points by category, and publishing, embedding or exporting the result.
Environ 7 minutesThis guide covers how to create a scatter plot in PlotSet from your own numbers, not a demo dataset: two numeric columns bound to X and Y, an optional category column colouring the points into groups, and axes and labels that actually say what the numbers mean. By the end you will have gone through the same click path a published PlotSet chart takes, from a raw spreadsheet to something you can embed or drop into a deck.
Seven minutes covers it when your sheet already has two clean numeric columns, most of that spent on labelling and styling rather than wrestling the data itself. Pulling numbers out of a wider report, or cleaning up a category column for grouping, adds another five to ten.
One row per record, with two numeric columns for the axes plus an optional label column and an optional category column for colour grouping. Just one header row, no merged cells.
| Campaign | Channel | Spend | Revenue |
|---|---|---|---|
| Spring Launch | Paid Search | 4200 | 18900 |
| Summer Sale | Paid Search | 6100 | 24300 |
| Back to School | Paid Social | 3800 | 11200 |
| Holiday Push | Paid Social | 7200 | 19800 |
| Retargeting Q1 | 1200 | 9800 | |
| Newsletter Promo | 900 | 6200 | |
| Influencer Collab | Paid Social | 5400 | 15100 |
| Display Retarget | Paid Search | 2600 | 10400 |
Both axis columns need to be numbers, not text that merely looks numeric. A currency symbol or a thousands separator left in as text will stop the column from binding to an axis, so strip formatting before uploading.
The label column (here, Campaign) is optional but worth keeping: it is what shows up in the tooltip and in any point labels you turn on later, instead of a bare coordinate.
The category column (here, Channel) is optional too. Skip it for a plain single-colour scatter plot, or keep it if you want the points grouped and coloured by that channel, region or segment.
One row per record, not per aggregate. If your source is already a pivot table of averages, decide whether you actually want the underlying rows plotted or a smaller, pre-aggregated table instead.
Le parcours exact dans PlotSet, d'un modèle vierge à un graphique terminé.
From the PlotSet dashboard, start a new project and choose the Scatter Plot template. It loads as a scatter plot with groups and comes with sample data already bound, so you can see the exact column layout it expects before touching your own file.
Put your two numeric measures in their own columns, add a label column if you want readable tooltips, and add a category column only if you plan to colour the points by group. Drop any subtotal or average row before uploading.
Import a CSV or .xlsx file from the data panel, paste rows straight from a spreadsheet, or connect a Google Sheet if you want later edits to flow through without a re-upload. PlotSet reads the first row as headers automatically.
Assign one numeric column to the X axis and the other to Y. PlotSet usually guesses correctly when your headers are clear, but check the mapping manually if your sheet carries extra numeric columns it might have picked the wrong one from.
If you included a category column, bind it to colour so each group gets its own shade and its own entry in the legend. Skip this step entirely for a single-series scatter plot: one colour is correct when there is nothing to group.
Name both axes with real units, not just the column header. Turn on point labels for the handful of records that matter, usually the outliers, and leave the rest to the hover tooltip so the chart does not turn into a wall of text.
Reduce mark size or add transparency if points are stacking into a solid blob. Only turn on the fitted trend line once you have actually looked at the shape of the cloud, since a straight line through a curved or clustered relationship reports a pattern that is not there.
Publish for a live interactive chart with hover tooltips and a shareable link, copy the embed code to drop it into a page, or export a PNG, SVG or PDF if you need a static image for a deck or report.
Partez du modèle Scatter Plot avec des données d'exemple déjà chargées, puis remplacez-les par les vôtres.
Ouvrir le modèleLa dernière ligne droite est la même pour tous les types de graphique : 'Share' transforme le graphique en lien public actif et en code d'intégration, 'Export' l'enregistre sous forme de fichier statique.
Activer le code d'intégration et le lien public revient à publier le graphique. Le lien peut être envoyé à qui vous voulez sans risque, et le code d'intégration s'insère dans n'importe quelle page sous la forme d'une iframe qui conserve les infobulles au survol et l'interactivité. Si vous modifiez le graphique par la suite, cliquez sur 'Recreate' pour répercuter ces changements sur le lien comme sur le code.
Le bouton 'Export' génère une copie statique à la taille dont vous avez besoin. Le SVG reste net à toutes les échelles et peut encore être modifié dans un outil de design ; PNG et JPG sont les choix sûrs pour les présentations et les documents. C'est le multiplicateur d'échelle qui vous donne une image nette sur un écran retina.
Past that, the legend and the eye both lose track of which shade is which. If your category column has more groups than that, filter to the ones that matter for the story or fall back to a single colour and call out one group with a label instead.
If the cloud reads as one dark shape, the fix is almost never a bigger chart. Drop the mark opacity to around 50 to 70 percent, or shrink the point size, so overlapping records show up as darker patches instead of one flat blob.
Do not zoom the axis in to make a weak relationship look tighter. Start from the full range of your data and only narrow it if you are explicitly calling out a subset, with a caption saying so.
The point worth a label is usually the one that breaks the pattern, not every point in a given group. Turn on manual point labels and pick the two or three records that carry the story, leaving the rest to the tooltip.
This usually means the category column was read as a number rather than text, so PlotSet tried to bind it to size or a colour scale instead of discrete groups. Check the column type in the data panel and force it to text if a category name looks like a code or an ID.
This is overplotting: too many points landing on top of each other at the resolution you are viewing. Lower the mark opacity, shrink the point size, or filter to a date range or subset before assuming the two variables have no relationship at all.
A linear trend line forces a straight line through data that might be curved, clustered into two groups, or dominated by one outlier. Turn the trend line off, look at the raw shape of the points, and only re-enable it if a straight line is genuinely the right summary.
Vous n'êtes pas sûr que le Scatter Plot soit le bon choix ? Découvrez ce que c'est, quand l'utiliser et quand l'éviter.
Two numeric columns, one per axis, with one header row. A label column and a category column are optional additions that drive the tooltip text and the colour grouping. CSV and .xlsx both work, and a pasted range from any spreadsheet tool is fine too.
Gratuit pour commencer avec plus de 300 modèles. Aucune carte bancaire requise.
