DVT UNIT-V

1. Multiple Quantities

📍 Scatterplots

 Definition:
A scatterplot is a graph that shows the relationship between two numbers by using dots.
Each dot shows one observation or item.

 Purpose:
Used to find out if two things are related, for example, does more sales mean more profit?

 Important Principles (Ben Jones):

o Know Your Goal:

Are you checking for correlation, finding outliers, or checking impact of another
factor?

o Use Correct Visualization:

Scatterplots are only for two numbers (quantitative data).

o Clarity & Aesthetics:

 Label X and Y axes properly with units.

 No need to always start axes at zero if it hides useful detail, but indicate it
clearly.

 Make sure dots are visible; avoid too many overlapping dots:

 Make dots smaller.

 Add transparency (alpha).

 Use outlines.

 Sample fewer points if needed.

 Use jitter (small random movement).

o Using Color, Size, Shape:

Add a third variable (like Region or Discount) by changing color or size of dots.

o Annotations:
Mark special points, like biggest sale, or profit loss.

o Check the Output:

Make sure your scatterplot clearly shows the pattern or the message you want.

 Reading a Scatterplot:
 o Positive Correlation: Dots rise from left to right.

o Negative Correlation: Dots fall from left to right.

o No Correlation: Dots are randomly scattered.

o Tight or Loose Clusters:

Tight = Strong relation; spread out = Weak relation.

o Shape:
Straight line = Linear; Curved = Non-linear.

o Outliers:
Dots far away from others — special cases needing investigation.

 Warning:
Scatterplots can only show correlation — they do not prove one thing causes the other.

 Example:

o Sales vs. Profit scatterplot.

o Color dots by Product Category.

o Add a horizontal line at Profit = 0 to separate profits and losses.

📍 Stacked Bars

 Definition:
A stacked bar chart is a bar chart divided into segments, each showing part of the total.

 Purpose:
To compare totals and also the components inside each total.

 Types:

o Standard Stacked Bar:

Bars have different lengths based on total value.

o 100% Stacked Bar:

All bars are same size, but segments show % share.

 Important Principles (Ben Jones):

o Know Your Goal:

Are you comparing totals or compositions (parts)?

o Use Correct Visualization:

If comparing parts, stacked bar is good; if only totals, simple bar may be better.

o Design for Clarity:

 Label each part clearly.

 Choose different colors for different parts.

  Always stack segments in the same order for all bars.

o Limitation:
It’s harder to compare the middle/top parts across bars (bottom part is easiest to
compare).

 Example:

o Sales in different regions stacked by product category (Furniture, Office Supplies,

Technology).

📍 Regression and Trend Lines

 Definition:

o A Trend Line shows the overall direction of data points.

o A Regression Line is the best fit line mathematically calculated.

 Purpose:

o To find and explain trends.

o To check if the relation is positive, negative, or neutral.

 Key Points:

o Slope (m): How fast Y changes when X changes.

o Intercept (c): Y value when X = 0.

o R² (R-squared): Percentage of how well the trend line fits the data (0%–100%).

o P-value: Tells if the trend is statistically important (<0.05 is good).

 Principles:

o Add trend line only if it helps understanding.

o Choose right model (linear or curved).

o Always show the equation and statistics if possible.

o Trend line should not hide data points.

 Example:

o Trend Line on scatterplot of Sales vs. Profit:

 Positive slope → Higher sales = Higher profits (generally).

 Moderate R² → Other factors also affect profit.

📍 Quadrant Chart
  Definition:
A scatterplot with two crossing reference lines that divide the plot into 4 sections
(quadrants).

 Purpose:
To categorize observations into groups for better strategic decisions.

 Key Points:

o Choose important X and Y variables.

o Choose meaningful reference lines (mean, median, zero, or benchmark).

o Label quadrants clearly.

o Explain how reference lines were set.

 Quadrant Interpretation:

o Top-Right: High on both axes → Best performers.

o Top-Left: High Y, Low X → Good profit but low sales.

o Bottom-Left: Low on both → Poor performance.

o Bottom-Right: High sales, low profit → Needs attention.

 Example:

o Quantity Sold vs. Profit Ratio for Products.

o Top-Right quadrant → "Star Products."

2. Changes Over Time

📍 Line Chart

 Definition:
A chart connecting data points over time using lines.

 Purpose:
To show trends, patterns, seasonality, and overall movement across time.

 Construction:

o X-axis: Time (continuous: Year, Quarter, Month).

o Y-axis: Measures like Sales, Profit.

 Interpretation:

o Upward slope = Increasing trend.

o Downward slope = Decreasing trend.

o Peaks and valleys = Seasonal variations.

  Tips:

o No need for zero baseline always.

o Annotate important events.

 Example:

o Monthly Sales Trend.

📍 Dual-Axis Line Chart

 Definition:
Line chart with two Y-axes showing two different measures.

 Purpose:
To compare two trends together over time.

 Caution:

o Scaling of axes may mislead.

o Careful labeling and explaining needed.

 Example:

o Sales vs. Quantity Sold over years.

📍 Connected Scatterplot

 Definition:
A scatterplot where points are connected based on time order.

 Purpose:
To show both the relationship between two variables and how it changes over time.

 Steps:

o Plot Sales vs. Profit.

o Connect points by Date.

 Tip:

o Keep it simple with fewer points.

📍 Date Field Type and Seasonality

 Understanding Date Field:

o Continuous Dates: Smooth time flow (good for trends).

o Discrete Dates: Separate categories (good for comparison).

  Finding Seasonality:

o Check for repeating patterns.

o Add Moving Average to smooth fluctuations.

 Example:

o Observe Sales patterns across months or years.

📍 Timeline

 Definition:
A graphical representation of events or changes in order.

 Purpose:
To highlight key milestones or important periods.

 Steps:

o Date on X-axis.

o Sales or Profit on Y-axis.

o Add annotations for key events.

📍 Slope Graph

 Definition:
A simple graph to compare two time points for multiple categories using sloped lines.

 Purpose:
To show increase or decrease across time for categories.

 Steps:

o Plot Year (start and end) on X-axis.

o Plot Measure (Sales, Profit) on Y-axis.

o Draw lines for each Category or Region.

o Label start and end points.

3. Maps and Locations

📍 Circle Maps

 Definition:
Maps with circles sized by a measure (e.g., Sales).

 Purpose:
 o Compare data across locations.

 Example:

o State-wise Sales shown with different sized circles.

 Tips:

o Adjust circle size to avoid overlap.

o Add labels and interactive tooltips.

📍 Filled Maps

 Definition:
Maps where areas (states, countries) are shaded by a value (like Sales).

 Purpose:

o Spot patterns geographically.

 Example:

o State-wise Sales colored (dark blue = high, light blue = low).

 Tips:

o Use appropriate color ranges.

o Use maps only if geography matters.

📍 Dual-Encoded Maps

 Definition:
Map using two visual encodings together: Size + Color.

 Purpose:

o Show two measures at once (e.g., Sales and Profit).

 Example:

o Size = Sales volume, Color = Profitability.

 Tips:

o Keep encodings clear.

o Add tooltips and legends.