Use Power BI Report Skills and
Claude Code, to create a complete Dashboard.

Visualizations

• Six KPI cards cardVisual Average high and low, total precipitation, snow days, and record high/low. Each carries a colored left accent bar assigned by meaning - rose for heat, cyan for cold, blue for water, violet for snow - so the strip is readable at a glance rather than being six identical tiles.
• Yearly Temperature Anomaly lineChart Each year's mean plotted as a departure from the fixed 1940–1980 baseline, with a reference line at zero. Native anomaly detection is enabled, so statistically unusual years are flagged automatically with explanations on hover - something the JavaScript build has no equivalent for.
• Average Temperature by Month areaChart High, mean and low across the calendar year, layered to show the seasonal envelope and how wide the daily range runs in each month.
• Precipitation by Month columnChart Rain and snow stacked per month, so the winter snow contribution reads as a distinct violet band on top of the rainfall total.
• Share of Precipitation by Season donutChart The direct successor to the original's hand-rolled D3 donut. Summer leads at 27.06%, and the four seasons sit within about 2.5 points of each other - Charlotte has no dry season.

Visualizations

• Yearly Average High, Mean and Low lineChart Three series across 86 years with a dashed baseline reference. All three lines drift upward, and the low-temperature line rises most - overnight lows are warming faster than daytime highs. 2026 is filtered out; a 134-day partial year would plot as a false cliff.
• Daily High vs. Low, Colored by Season scatterChart Every one of the 31,545 days as a single point, colored by season. The diagonal spread is the daily temperature range, and the four seasonal clusters separate cleanly along it without any need for a legend lookup.
• Days per Temperature Band clusteredBarChart How often Charlotte reaches each daily-high range, from Freezing to Extreme Heat. Built on a banding column in the model, so the same six categories drive the machine-learning visual on the Precipitation page.
• Hottest Days on Record tableEx The top of the record, ranked. Column totals are explicitly disabled - a "total" row summing record temperatures would be meaningless. Right-click any row to drill through to that day's full detail.

Visualizations

• Rolling 12-Month Precipitation lineChart Every 12-month total since 1940 against the 1991–2020 normal, drawn as a dashed amber reference. Native forecasting projects ahead with a 95% confidence band. The plunge at the right-hand edge is the current deficit - the argument the Drought page then makes in full.
• Monthly Precipitation by Decade columnChart Each month broken into its nine decades, so both the seasonal shape and the decade-to-decade variation are visible in one chart.
• Average Precipitation by Month treemap Successor to the original's Plotly treemap. Area is proportional to the monthly mean and a single-hue cyan gradient reinforces it - a categorical palette here would imply difference where the data only has magnitude.
• Every Year as a Bubble scatterChart Three dimensions at once: mean temperature on x, total precipitation on y, and snow days as bubble size. The lone outlier at the bottom left is a partial year.
• What Makes a Heavy-Precipitation Day More Likely? keyDriversVisual Key Influencers runs machine learning inside the visual, ranking the conditions associated with days above 1.4 inches - the 95th percentile of wet days. It reports that Extreme Heat raises the likelihood of a dry day by 1.13×, with October and Fall close behind. There is no JavaScript equivalent; this is the clearest case of Power BI doing something the original build simply cannot.

Visualizations

• Precipitation Heat Grid pivotTable Every month since 1940 as one cell - the densest visual in the report, roughly 1,030 data points in a single view. Conditional background formatting runs dark for dry and cyan for wet. The blank 1942–43 rows are visible as genuine gaps, which is precisely the honesty the zero-filling original loses.
• Decompose Precipitation decompositionTreeVisual An interactive breakdown the viewer drives: click any + to branch by period, season, intensity, decade, month or day class. AI splits are enabled, so choosing "High value" or "Low value" lets Power BI pick the most interesting branch itself. This is genuinely impossible in the static JavaScript build.
• Season × Decade Mean Temperature pivotTable Nine decades against four seasons with diverging conditional formatting. Warming reads left-to-right down the grid: summer climbs from 77.6°F in the 1940s to 79.6°F in the 2020s, and winter from 42.6°F to 45.9°F - winters are warming faster than summers.

Visualizations

• Alert banner textbox + DAX Not static text - a DAX measure assembles the whole sentence from live values, so the inches, the deficit, the percentage and the classification all update together.
• Six drought KPIs cardVisual Trailing 12-month actual, the normal, the deficit, percent of normal, the current dry streak, and where this window ranks historically. Amber for actual, cyan for normal, rose for deficit.
• Rolling % of Normal vs. NOAA Bands lineChart The centerpiece, and the one visual that had to be reinvented. The original used a hand-drawn SVG gauge with D0–D4 severity bands; Power BI's native gauge has no concept of classification bands. The replacement encodes all five NOAA categories as independently colored shaded zones behind a rolling percent-of-normal series - which carries strictly more information than the gauge did: current severity, and how it developed, and how it compares to every past drought.
• NOAA Drought Classification textbox grid The D0-D4 scale spelled out beside the chart, mapping each severity band to its percent-of-normal range. Power BI has no legend visual for a custom classification, so this is a hand-built grid of 16 aligned textboxes - the single largest piece of page furniture in the report, and the reason this page carries 39 objects for 5 charts. Below it, a DAX measure writes the live ranking sentence: the current window is 22nd driest of 989 complete 12-month windows since 1940, drier than 98% of the record.
• Last 12 Months vs. Normal clusteredColumnChart Month-by-month actual against the 1991–2020 normal. Clustered rather than stacked, because stacking hides exactly the comparison being made.
• The Ten Driest 12-Month Periods tableEx Historical context, ranked. Overlapping windows are expected and labeled as such - 2001–02 and 2007–08 each appear more than once because those droughts persisted across consecutive months.

Visualizations

• Six context KPIs cardVisual Observation days, average high and low, total precipitation, wet days and the longest dry streak - all recalculated for whatever slice you drilled from, so the summary always describes the rows below it.
• Daily Observations tableEx The underlying grain, fully exposed: date, season, day name, high, low, mean, precipitation, snowfall and day class. This is the bottom of the model - the actual NOAA rows behind every aggregate in the report.

Why This Page Matters

Drill-through is the capability with no JavaScript counterpart in the original build. Every aggregate in the report is a summary of specific days, and this page makes those days reachable in two clicks from anywhere - without a single line of routing code, and without ever loading 31,545 rows into a browser.