Linear transformation: Adding or subtracting a constant to each value in a dataset.
Translation: Moving all data points by the same amount without changing their relative positions.
- Linear transformation: x → x + c (where c is a constant)
- Mean: New mean = Original mean + c
- Standard deviation: New std dev = Original std dev (unchanged)
Each value is increased by 15
This is a linear transformation: x → x + 15
Original mean = 50
New mean = Original mean + 15 = 50 + 15 = 65
Original standard deviation = 10
Linear transformations do not affect spread measures
New standard deviation = 10 (unchanged)
Standard deviation measures spread: σ = √[Σ(xi - μ)²/n]
After transformation: σ_new = √[Σ((xi + 15) - (μ + 15))²/n] = √[Σ(xi - μ)²/n] = σ
The constant cancels out, so standard deviation is unchanged
New mean = 65, New standard deviation = 10
The new mean is 65 and the new standard deviation is 10. When we add a constant to each value, the mean increases by that constant, but the standard deviation remains unchanged because the relative distances between values stay the same.
• Linear transformation: x → x + c
• Mean transformation: μ_new = μ_orig + c
• Spread invariance: σ_new = σ_orig
Scaling transformation: Multiplying or dividing each value in a dataset by a constant.
Proportional change: All values change by the same multiplicative factor.
Each value is multiplied by 3
This is a scaling transformation: x → 3x
Original mean = 20
New mean = Original mean × 3 = 20 × 3 = 60
Original median = 18
New median = Original median × 3 = 18 × 3 = 54
Original standard deviation = 5
Scaling transformations affect spread measures by the same factor
New standard deviation = 5 × 3 = 15
For standard deviation: If each value is multiplied by 3, the distances from the mean are also multiplied by 3
σ_new = √[Σ(3xi - 3μ)²/n] = √[Σ9(xi - μ)²/n] = 3√[Σ(xi - μ)²/n] = 3σ
The new mean is 60, the new median is 54, and the new standard deviation is 15. When we multiply each value by 3, all measures (both center and spread) are multiplied by 3. The relative positions of values are preserved, but all distances are scaled by the same factor.
• Scaling transformation: x → cx
• Center scaling: Mean_new = Mean_orig × c
• Spread scaling: σ_new = σ_orig × c
Combined transformation: A transformation that includes both scaling and linear components (y = ax + b).
Sequential effects: Scaling affects both center and spread, while linear transformation affects only center.
Transformation: y = 2x + 10
This consists of: (1) Scaling by 2, then (2) Adding 10
After scaling by 2:
New mean = 40 × 2 = 80
New median = 35 × 2 = 70
New standard deviation = 8 × 2 = 16
After adding 10 to the scaled values:
New mean = 80 + 10 = 90
New median = 70 + 10 = 80
New standard deviation = 16 + 0 = 16 (unchanged by linear transformation)
For transformation y = ax + b:
Mean_new = a × Mean_orig + b
Median_new = a × Median_orig + b
σ_new = |a| × σ_orig
Using a = 2, b = 10: Mean = 2×40 + 10 = 90, Median = 2×35 + 10 = 80, σ = 2×8 = 16
Mean_new = 90, Median_new = 80, Standard deviation_new = 16
These match our sequential calculation
The new mean is 90, the new median is 80, and the new standard deviation is 16. For the transformation y = 2x + 10, both center and spread measures are affected by the scaling factor (2), while only center measures are affected by the additive constant (10).
• Combined transformation: y = ax + b
• Center transformation: Center_new = a × Center_orig + b
• Spread transformation: Spread_new = |a| × Spread_orig
SD → σ
SD → |c|σ
SD → |a|σ
Linear transformation: Adding or subtracting a constant to each value (x → x + c).
Scaling transformation: Multiplying or dividing each value by a constant (x → cx).
Combined transformation: A combination of scaling and linear transformations (x → ax + b).
Measure invariance: How certain measures respond to transformations.
- Linear transformations: Affect center measures, leave spread unchanged
- Scaling transformations: Affect both center and spread measures
- Combined transformations: Scale both center and spread, then shift center
• Linear (x+c): Center changes by c, spread unchanged
• Scaling (cx): Center and spread both multiplied by |c|
• Combined (ax+b): Center → a×center+b, spread → |a|×spread
Scale conversion: Changing the unit of measurement while preserving the relative relationships between values.
Proportional scaling: Maintaining the same relative distances between values after transformation.
Formula: New GPA = 25 × Old GPA
This is a scaling transformation with factor a = 25
Original: mean = 3.2, standard deviation = 0.5
New mean = 25 × Original mean = 25 × 3.2 = 80
New standard deviation = 25 × Original standard deviation = 25 × 0.5 = 12.5
Original scale: Mean = 3.2, SD = 0.5 (on 4.0 scale)
New scale: Mean = 80, SD = 12.5 (on 100-point scale)
The transformation preserves the relative positions of students
A student with original GPA of 3.2 (mean) now has 80 (new mean)
A student one standard deviation above the mean: 3.2 + 0.5 = 3.7 → 25 × 3.7 = 92.5
One standard deviation above new mean: 80 + 12.5 = 92.5 ✓
The new mean is 80 and the new standard deviation is 12.5. The conversion from 4.0 scale to 100-point scale multiplies all statistics by 25. The mean GPA of 3.2 becomes 80, and the standard deviation of 0.5 becomes 12.5. The relative positions of students remain the same, just on a different scale.
• Scaling factor: a = 25
• Center scaling: Mean_new = 25 × Mean_orig
• Spread scaling: σ_new = 25 × σ_orig
Temperature conversion: A real-world application of linear transformations with both scaling and shifting components.
Unit conversion: Changing measurement units while preserving physical relationships.
°C = (°F - 32) × 5/9
°C = (5/9) × °F - (5/9) × 32
°C = (5/9) × °F - 17.78
This is in the form y = ax + b where a = 5/9 and b = -17.78
Original mean = 68°F
New mean = a × Original mean + b = (5/9) × 68 + (-17.78)
New mean = 37.78 - 17.78 = 20°C
Original standard deviation = 10°F
Standard deviation is affected only by the scaling factor (a)
New standard deviation = |a| × Original standard deviation = (5/9) × 10 = 5.56°C
Using the formula: New mean = (68 - 32) × 5/9 = 36 × 5/9 = 20°C ✓
For standard deviation: The linear shift (-32) doesn't affect spread
Only the scaling by 5/9 affects the standard deviation
The mean temperature of 68°F becomes 20°C
The standard deviation of 10°F becomes 5.56°C
The variability is reduced because the Celsius scale has smaller units
The new mean is 20°C and the new standard deviation is 5.56°C. The conversion from Fahrenheit to Celsius involves both scaling (by 5/9) and shifting (subtracting 32 then scaling). The mean is affected by both operations, but the standard deviation is only affected by the scaling operation, reducing the variability from 10°F to 5.56°C.
• Combined transformation: y = ax + b
• Center transformation: Both scaling and shifting apply
• Spread transformation: Only scaling applies (shift doesn't affect spread)
Transformation: A mathematical operation applied to each value in a dataset.
Linear transformation: Adding or subtracting a constant (translation).
Scaling transformation: Multiplying or dividing by a constant (expansion/compression).
Combined transformation: Both scaling and linear operations applied together.
Measure invariance: How statistical measures respond to transformations.
- Identify transformation type: Linear, scaling, or combined
- Apply to center measures: Mean, median, mode
- Apply to spread measures: Standard deviation, variance, range, IQR
- Verify with examples: Check with sample values
• Linear (x+c): Center measures shift by c, spread measures unchanged
• Scaling (cx): All measures scale by |c| (except variance by c²)
• Combined (ax+b): Center → a×center+b, Spread → |a|×spread
• Variance scaling: Var(ax+b) = a²×Var(x)
Linear: Shifts center only, preserves spread
Scaling: Affects both center and spread
Combined: Both effects applied together
Invariance: Some relationships preserved under transformation
Real-world: Practical applications in unit conversions
Analysis: Transformations affect center and spread measures differently depending on the type of transformation applied.
- Linear transformations: Only affect center measures
- Scaling transformations: Affect both center and spread
- Combined transformations: Apply both effects sequentially
- Shape preservation: Distribution shape remains the same
- Unit conversion: Maintain relative relationships between values