Correlation & Regression: Types, Calculation & Business Use

TAKEAWAYS:

• Correlation measures how two variables move together (positive/negative/none), but does not prove causation.
• Regression predicts one variable’s value based on another using the least squares method (best-fit line).
• Pearson’s r (–1 to +1) quantifies correlation strength; r² explains how much variation is explained by the relationship.
• Linear regression uses the formula $y=a+bx$ to model trends in business data (sales vs. ads, production vs. cost).
• Scatter plots visually show correlation before calculating numbers.
• NEB exams test: interpreting graphs, calculating r and regression line, and applying concepts to business scenarios.

---

1. What is Correlation?

Correlation studies how two variables change together. For example:

• Sales vs. Advertising: If sales increase as ads increase, they are positively correlated.
• Temperature vs. Ice Cream Sales: Higher temperature → more sales (positive).
• Study Hours vs. Exam Failure: More study → fewer failures (negative).

Types of Correlation

• Perfect Positive (+1): Points lie exactly on a straight line upward.
• Strong Positive (0.7–1): Clear upward trend.
• Weak Positive (0.3–0.7): Slight upward trend.
• No Correlation (0): Points scattered randomly.
• Weak/Negative Correlation (–0.3 to –0.7): Slight downward trend.
• Perfect Negative (–1): Points lie exactly on a straight line downward.

How to Calculate Correlation Coefficient (r)?

The formula for Pearson’s r (most common in NEB) is: $$r=\frac{n(\sum XY)-(\sum X)(\sum Y)}{\sqrt{[n\sum X^2-(\sum X{)}^2][n\sum Y^2-(\sum Y{)}^2]}}$$ Where:

• $n$ = number of pairs
• $\sum XY$ = sum of $X\times Y$ for all pairs
• $\sum X$, $\sum Y$ = sum of X and Y values
• $\sum X^2$, $\sum Y^2$ = sum of squares of X and Y

---

Worked Example 1: Calculate r for Sales vs. Ads

Step-by-Step Calculation:

1. $n=4$
2. $\sum X=20$, $\sum Y=105$
3. $\sum XY=660$, $\sum X^2=120$, $\sum Y^2=3675$
4. Plug into formula: $$r=\frac{4(660)-(20)(105)}{\sqrt{[4(120)-20^2][4(3675)-105^2]}}$$
5. Numerator: $2640-2100=540$
6. Denominator: $$\sqrt{(480-400)(14700-11025)}=\sqrt{(80)(3675)}=\sqrt{294000}\approx 542.22$$
7. $r=\frac{540}{542.22}\approx 0.996$

Interpretation: Almost perfect positive correlation (0.996 ≈ +1).

---

2. What is Regression?

Regression finds the best-fit line (equation) to predict one variable from another. The most common is linear regression: $$y=a+bx$$

• $y$ = dependent variable (predicted)
• $x$ = independent variable (known)
• $a$ = y-intercept
• $b$ = slope (change in $y$ per unit $x$)

How to Find the Regression Line?

The formulas for $a$ and $b$ are: $$b=\frac{n(\sum XY)-(\sum X)(\sum Y)}{n\sum X^2-(\sum X{)}^2}$$ $$a=\frac{\sum Y-b\sum X}{n}$$

---

Worked Example 2: Find Regression Line for Ads vs. Sales Using the same data as above:

1. We already have:
   • $n=4$, $\sum X=20$, $\sum Y=105$, $\sum XY=660$, $\sum X^2=120$
2. Calculate $b$: $$b=\frac{4(660)-(20)(105)}{4(120)-20^2}=\frac{2640-2100}{480-400}=\frac{540}{80}=6.75$$
3. Calculate $a$: $$a=\frac{105-6.75(20)}{4}=\frac{105-135}{4}=\frac{-30}{4}=-7.5$$
4. Regression equation: $$y=-7.5+6.75x$$

Prediction: If ads ($x$) = 5, then predicted sales ($y$) = $-7.5+6.75(5)=26.25$.

---

3. Scatter Plots and Regression Lines

Key Observations:

• The line passes closest to all points (least squares method).
• Slope (6.75): For every 1 unit increase in ads, sales increase by 6.75 units.
• Intercept (–7.5): If ads = 0, sales would be –7.5 (not realistic; extrapolation risk!).

---

4. Difference Between Correlation and Regression

---

5. Applications in Business

1. Sales Forecasting: Predict future sales based on past trends.
2. Cost Estimation: Estimate production costs based on units made.
3. Marketing: Determine how much to spend on ads to achieve sales targets.
4. Economics: Study relationships between GDP, inflation, and unemployment.
5. Quality Control: Detect defects based on production variables.

---

6. Limitations

• Correlation ≠ Causation: Just because two variables move together doesn’t mean one causes the other.
  • Example: Ice cream sales and drowning incidents both rise in summer, but ice cream doesn’t cause drowning!
• Linearity Assumption: Regression assumes a straight-line relationship. If data is curved, it fails.
• Outliers: A single extreme value can distort the regression line.
• Extrapolation Risk: Predicting far beyond the data range is unreliable.

---

7. NEB-Style Questions (Practice!)

Short Answer (3–5 marks)

1. Define correlation coefficient. How does $r=0.8$ differ from $r=-0.5$?
2. What is the least squares method in regression?
3. Give two business scenarios where regression can be applied.

Calculation (7–10 marks)

4. Calculate the correlation coefficient for the following data:

   X	Y
   3	5
   5	7
   7	9
   9	11

5. Find the regression line for the data in Q4. Predict $Y$ when $X=6$.

Interpretation (5–7 marks)

6. The correlation between employee training hours ($X$) and productivity ($Y$) is $r=0.65$. Explain what this means for a manager.
7. A company finds that its regression equation for sales ($Y$) vs. ads ($X$) is $Y=10+2X$. If the company increases ads by 5 units, by how much will sales increase?

---

Exam Tip

1. Graphs First: Always plot a scatter plot before calculating r or regression. NEB often gives partial marks for correct graphs.
2. Units Matter: In regression, interpret the slope in context (e.g., "sales increase by ₹500 per ₹1000 spent on ads").
3. Sign of r: Positive/negative correlation depends on whether variables increase/decrease together.
4. Formula Recall: Memorize the r and regression formulas—they are tested in calculations.
5. Real-World Link: NEB loves questions like:
   • "A shopkeeper records sales vs. price. If $r=-0.9$, what does this imply?"
   • "Predict next month’s sales using the regression line."
6. Avoid Common Mistakes:
   • Forgetting to square $X$ and $Y$ in the denominator of $r$.
   • Misplacing decimals in calculations (always double-check!).
   • Assuming causation from correlation.

---

Final Note: Correlation and regression are powerful tools for business decisions. Master the formulas, practice with real data, and always interpret results in context!