Visualize sankey chart: easy guide, free example
Explain how energy sources feed power and split into residential and industrial use, checking balance at the shared node.
Read the guideEasy chart guides, free CSV examples and real Datamon visualizations. Explore charts, maps, analysis, styling, exports and specific data problems.
Explain how energy sources feed power and split into residential and industrial use, checking balance at the shared node.
Read the guideCompare a profile across several measures while checking the independent normalization used on each radial axis.
Compare nonnegative category values through sector areas with equal angles, rather than treating radius as value.
Compare category values as concentric arcs and keep the shared scale explicit.
Explore an ordered cycle, checking that the final-to-first connection has a meaningful interpretation.
Show a cyclic measure with a filled radial outline while disclosing that area can exaggerate visual differences.
Follow parent-to-child business revenue across rings, checking that hierarchy levels are not counted twice.
Read a revenue hierarchy from parent blocks to child blocks using nested rectangular spans.
Explore hierarchical revenue groups through nested circles, using the table when precise ranking matters.
Present survey shares as a finite grid and disclose the rounding needed to turn shares into cells.
Represent an aggregate share with repeated units while making clear that marks are not individual respondent records.
Compare market revenue through proportional areas rather than assigning revenue directly to radius or diameter.
Use explicit label weights for a compact overview; this example does not automatically extract words from documents.
Explore changing component patterns on a shifted baseline, then use another view for precise level comparisons.
Show additive components over time only when all series share a unit and their sum is meaningful.
Present an ordered trend with interpolation, explaining that a smooth curve does not add new observations.
Show values that remain effective until the next recorded change rather than implying continuous movement.
Connect two comparable measures for each category and inspect the direction and size of the endpoint gap.
Track changes in rank through ordered observations without mistaking a rank shift for a large value change.
Explore multivariate country profiles after checking each numeric axis and the effects of independent scaling.
Visualize two supplied bounds across categories without calling the band a confidence interval unless the method supports it.
Display supplied lower and upper bounds as capped intervals; the renderer does not estimate uncertainty from raw observations.
Compare lower and upper values for each category without adding the two endpoints as if they were components.
Show open, high, low and close in the required order and keep the reporting interval explicit.
Compare ordered stage values in a tapered layout without assuming every stage is a conserved flow.
Present a few part-to-whole survey shares in a half-circle while retaining the numeric percentages.
Compare one measure to its intended target and explain the maximum before interpreting the dial.
Show actual-to-target progress while checking how zero targets and values above target are handled.
Compare category size and internal composition together, using a table to verify both dimensions.
Rank market revenue with a stem and endpoint, retaining a common zero baseline for length comparison.
Compare two district measures on opposing sides while making the sign and centerline convention explicit.
Compare group distributions through estimated density, checking sample size before treating the shape as stable.
Keep individual repeated observations visible by separating marks that would otherwise overlap.
Show individual observations along a group axis while checking whether identical values overlap.
Separate overlapping observations with positional jitter that is display displacement rather than measured variation.
Summarize dense paired observations in hexagonal bins and disclose how bin size changes the apparent pattern.
Connect paired observations only when row order represents a meaningful sequence, not alphabetical country order.
Inspect three positive components after choosing a meaningful common total; unrelated indicators do not form a valid composition.
Explore source-target connections and weights without interpreting layout distance as geographic or causal distance.
Read connections between ordered nodes and use link values rather than arc height for quantities.
Explore weighted relationships among groups while checking direction and repeated links in the source table.
Show transitions between named stages, ensuring that similarly named categories at different stages remain identifiable.
Explore an ordered process measure with displayed reference limits, checking how the implementation derives those limits.
Arrange a matrix around a circle when the category order is useful and a rectangular heatmap remains available for comparison.
Explore a long ordered sequence through color while keeping its time order and color scale explicit.
Compare regional intensity with a rate or clearly named total and review every unmatched geographic identifier.
Place stores or events using latitude and longitude, checking coordinate order against known locations.
Show a numeric measure at each store location while preserving a size legend and checking overlap.
Group nearby store points at the current map scale without presenting clusters as administrative regions.
Explore where recorded locations concentrate, keeping observation coverage separate from real-world risk.
Connect trade-route origins and destinations using a common flow unit and an explicit direction.
Compare two regional measures together only after explaining the two-dimensional color legend.
Assign regions to named categories without implying that categorical colors form a numeric ranking.
Rank regional revenue so readers can compare market totals from one zero baseline.
Give long country or department names enough space in a ranked comparison.
Compare a short list of market totals with vertical columns and readable labels.
Compare two measures within each district while keeping a shared baseline.
Show how components contribute to each district total before comparing the totals.
Trace solar, wind and hydropower observations in time order without confusing levels with growth rates.
Emphasize the level of a time series while checking whether overlapping fills hide another series.
Inspect the distribution of a numeric field before reporting its average or fitting a model.
Compare medians, spread and unusual observations across groups rather than relying on a single mean.
Show a small set of mutually exclusive survey shares that belong to one complete total.
Present survey composition when a compact part-to-whole view matters more than exact ranking.
Compare paired income and life-expectancy values while retaining the country behind every point.
Add population as a third numeric dimension without changing the income and life-expectancy coordinates.
Compare a district-by-measure matrix using one explained color scale and the original numbers.
Compare business-unit revenue shares while keeping parent groups and child categories distinguishable.
Explain how energy sources feed power and split into residential and industrial use, checking balance at the shared node.
Compare two district-level measures with adjacent columns, retaining the same unit for both series.
Show both district totals and the components making up each total in a compact vertical layout.
Compare relative district composition after normalizing each row to its own total.
Compare percentage composition vertically while using a separate table for absolute district totals.
Show revenue and EBITDA margin together while labeling the independent currency and percentage scales.
Reconcile opening revenue, signed changes and closing revenue without counting totals as another change.
Track ordered sales stages and calculate conversion only when the stages represent a comparable cohort.
Present one current measure with a defined period and comparison instead of hiding the denominator.
Compare agreement distributions using ordered response categories and a clearly defined neutral midpoint.
Show task start and end dates to reveal overlap without treating a schedule as proof of completion.
Compare actual performance against a common numeric scale with less ink than filled bars.
Show the gap between budget and actual for each category, keeping both endpoints visible.
Compare actual performance to target on the same scale with the target defined before judging attainment.
Find which revenue categories account for most of a total using descending values and cumulative share.
Inspect pairwise linear associations between numeric indicators after checking missing pairs and outliers.
Compare valuation intervals from different methods while keeping every method in the same currency and unit.
Keep exact category values available when readers need lookup, sorting and precise comparison.
Inspect valid count, missing values, mean and median before choosing a chart.
Compare each ordered value with its previous observation; irregular gaps are not automatically annualized.
Compare a dated monthly observation with the same month one year earlier, keeping unmatched months missing.
Compare adjacent calendar months rather than whichever row happens to precede the current row.
With monthly input, Datamon uses a three-month calendar lag; aggregate quarter totals first when that is the intended measure.
Summarize endpoints over an explicit elapsed span; the resulting constant rate does not describe each intermediate year.
Divide category totals by the same denominator and retain that denominator beside the percentages.
Sort a chosen measure with an explicit direction and limit; a selected top five is not the full distribution.
Smooth an ordered series with a stated trailing window and distinguish early partial windows from complete ones.
Flag observations outside IQR fences for review, keeping the original records until the cause is understood.
Calculate Pearson correlation and a linear fit on correctly paired numeric observations, then inspect the scatter plot.
Combine repeated categories with an explicit sum, making the observation unit clear before aggregating.
Compare actual and comparison fields as matched pairs; a positive cost variance can be unfavorable.
Compare actual with a defined target and keep a zero target undefined rather than showing an infinite achievement rate.
Rank category totals and show cumulative contribution without assuming an 80/20 pattern must appear.
Keep funnel stages in their operational order and compare stage volumes from the same reporting scope.
Compare relative changes after dividing every value by the same positive baseline within its series.
Accumulate flow observations such as revenue through time; adding price levels usually has no useful interpretation.
Calculate relative price changes in order and specify whether dividends, fees and currency effects are included.
Measure distance below the running peak, with zero meaning the series is at its recorded peak.
Decompose actual-minus-comparison changes by category and reconcile their sum without adding an invented opening balance.
Start a recurring report from a clean observation table instead of a formatted worksheet with merged cells. The free example lets you inspect the expected shape before uploading your own file.
A correct chart starts with columns assigned to the intended roles. A source column belongs in Source for a Sankey, while a numeric predictor belongs in X for a scatter plot.
Highlight the series a report discusses while keeping comparison series readable. Datamon supports individual series settings rather than requiring a new chart for every visual treatment.
Turn a chart into a figure that can stand on its own by naming the measure, observation period and source. Add an annotation only where it explains a specific mark or event.
Prepare a report figure at the dimensions where it will be read. Preview zoom changes the viewing scale, while canvas width and height determine the figure layout.
Add an image layer when a logo or contextual image helps identify the report. Keep image decoration separate from the marks that encode the data.
Deliver a chart in the format its destination needs. A report may need PDF, a web page PNG, and a design workflow SVG; the same project can produce separate outputs.
Share a public snapshot when readers need an interactive chart and its underlying values. Publishing and local file export are different delivery choices.
Keep datasets, visual settings and annotations together so a report figure can be revised instead of reconstructed. Account saving and a downloaded backup provide different recovery paths.
Use a deterministic calculation for numbers and an optional AI explanation for a draft interpretation. Review proposed chart edits against the current dataset before applying them.
Explain a sequence through supported motion modes, keeping time and category roles separate. A useful animation shows change that a static frame cannot show as clearly.
Choose region boundaries or coordinates according to what each record represents. A store belongs to a point; a provincial rate belongs to the corresponding administrative region.
Use ISO month values such as 2023-01 and sort them before importing.
Check the decimal separator before plotting rainfall.
Leave an unreported renewable-energy value blank.
Count earthquake event IDs within each UTC date before making daily columns.
Divide each country’s population by its own 2000 population and multiply by 100.
For a combined rate, estimate users as population × internet_pct / 100 for each country, add those estimates, and divide by the combined population.
Subtract the starting share from the ending share to calculate percentage-point change.
Make two separate rankings: internet-use percentage and estimated internet users.
Join the two indicators on country code and year, never on row position.
Precompute log10 of positive GDP per capita to spread out countries clustered at the lower end of the income scale.
Use population as the Size field while keeping income and life expectancy as the two positional measures.
Choose a common year before ranking countries.
Match the data’s ISO-style three-letter country codes to the boundary dataset’s identifiers before interpreting a choropleth.
Use the population share with access to electricity for a regional intensity comparison.
Use the same numeric color domain when comparing internet-use maps for 2000 and 2023.
Assign latitude and longitude to their named coordinate roles.
Use equal-size points for locations and explain magnitude separately.
A histogram of this dataset starts at magnitude 5 because that is the query threshold.
Use depth in kilometers as X and magnitude as Y, keeping both measurements from the same event.
Convert each event timestamp to the chosen time zone before extracting the date.
A temperature anomaly is a difference from a reference-period average.
Compute each trailing mean from the current year and previous four annual observations.
Use a linear value scale that includes the full negative and positive range.
Add daily precipitation to obtain monthly precipitation in millimeters.
Prepare one month column and one numeric temperature column per city.
Define a wet day before counting.
Use urban population share and its complement, 100 − urban share, as two components of one whole.
Set the final chart size before adjusting text.
Give an AI assistant the exact columns, year, source and allowed claims before asking for a chart explanation.
Publish the aggregated daily table when the story concerns daily counts.
Prepare a CSV, verify dates and missing values, and choose a chart that preserves the meaning of your observations.
Separate observations from spreadsheet presentation, validate the imported values, and build an interactive view without losing context.
Match the analytical question to a visual encoding: comparison, trend, distribution, relationship or geographic pattern.
Use bars for discrete comparisons and lines for ordered change; avoid misleading scales and accidental connections.
Choose readable labels, useful annotations and a format that holds up at the size your audience will see.
Separate deterministic calculations from AI interpretation, verify proposed changes, and control what data reaches a provider.
Match regional identifiers, normalize the measure, and make classification and missing data visible.
Compare prices, growth and indexed performance with clear baselines, honest scales and explicit missing-value handling.