title: “Climate Change Dataset: Flexdashboard Analysis” output: flexdashboard::flex_dashboard: orientation: rows vertical_layout: fill theme: cosmo source_code: embed ——————
This flexdashboard analyzes the Climate Change Dataset using cleaned monthly climate observations from 2020 to 2024. The dashboard emphasizes three goals:
Missing values were handled carefully in two steps. First, impossible or placeholder entries such as Unknown, NAN, and sentinel values like 99999 were converted to missing. Next, missing values were imputed using the mean within month, which is more defensible than a single grand mean because many climate variables have strong seasonal structure. When a month-level mean was still unavailable, the overall variable mean was used as a fallback.
This design follows common visualization principles from course-style data visualization readings: choose the graph type that matches the analytical goal, reduce clutter, label clearly, and avoid decorations that do not improve interpretation.
The time-series view is useful because climate questions are inherently temporal. A connected line reveals continuity and makes it easier to detect trends than isolated points. The faceted yearly comparison then separates multiple measures into smaller panels, preventing overplotting and making the dashboard easier to read.
A reasonable interpretation from this dataset is that CO2 concentration increases over the years, while other variables such as average temperature, humidity, and sea-surface temperature fluctuate more strongly across time. This does not prove causation, but it does show why a multi-variable dashboard is helpful: climate patterns rarely move in isolation.
The scatterplot was selected because both CO2 concentration and average temperature are continuous numeric variables. This makes it easier to inspect whether higher CO2 values tend to align with higher or lower temperatures in this sample. The regression line provides a compact visual summary of direction.
The monthly boxplot complements that relationship view by showing variation, not only averages. This matters because climate interpretation should consider spread and seasonality, not just mean values.
This section focuses on variables that often vary with seasonal weather conditions: precipitation, humidity, cloud cover, and solar irradiance. The bar chart is appropriate here because month is a discrete category, and the main task is comparison across categories rather than continuous trend estimation.
Together, the chart and table make it easier to discuss how seasonal conditions shift through the year and how those changes may align with changes in temperature and atmospheric measures.
The heatmap provides a compact overview of many pairwise relationships at once. This is useful in climate analysis because no single bivariate plot can summarize the whole system. Positive values indicate variables that move together, while negative values indicate inverse movement.
The heatmap also reflects a best-practice principle from data visualization readings: when comparing many variables simultaneously, use a layout that supports pattern recognition without forcing the audience to mentally combine multiple separate figures.
You can keep this section as written or replace it with the exact course readings assigned in your class.
This flexdashboard applies those ideas by:
To knit this dashboard successfully, keep
climate_change_dataset.csv in the same
folder as this .Rmd file, then click
Knit in RStudio.