Term 3 · Module 2 of 4

Simple Linear Regression

Business Statistics for Entrepreneurs III

Practise this module in Claude, ChatGPT or Gemini

Set it up free
Mind map

6.2.1 Linear Regression: Purpose, Language, and Model

Regression predicts a continuous numerical outcome from one or more explanatory variables. It is supervised learning: the training observations include both the predictor and the outcome. In simple linear regression (SLR), one predictor xx is used to explain or predict one response yy through a relationship linear in its unknown coefficients.

Galton studied heights in 905 adult children from 205 families, converting female heights to male equivalents by multiplying by 1.081.08. Children of above-average mid-parent height tended to be shorter than their parents, while children of below-average mid-parent height tended to be taller. He called this movement toward the offspring population mean regression.

Term Meaning
Dependent variable yy Outcome, response, or quantity being predicted
Independent variable xx Predictor, explanatory variable, or feature
Error εi\varepsilon_i Random unexplained part of observation ii
Regression coefficients β0,β1\beta_0,\beta_1 Population intercept and slope

The population model is

yi=β0+β1xi+εiy_i=\beta_0+\beta_1x_i+\varepsilon_i

The systematic component β0+β1xi\beta_0+\beta_1x_i is the conditional expected response at xix_i; εi\varepsilon_i captures what the predictor and line do not explain. Examples include sales from advertising, customer ratings from wait times, treatment cost from patient weight, and maintenance cost from bus age. With two or more predictors the model is multiple regression.

“Linear” refers to the coefficients, not necessarily to the shape in xx. Both yi=β0+β1xi2+εiy_i=\beta_0+\beta_1x_i^2+\varepsilon_i and yi=β0+β1log⁡(xi)+εiy_i=\beta_0+\beta_1\log(x_i)+\varepsilon_i are linear in the parameters. A form such as yi=β0+eβ1xi+εiy_i=\beta_0+e^{\beta_1x_i}+\varepsilon_i is nonlinear in a coefficient. Transformations can sometimes turn a nonlinear-looking relationship into a model linear in its parameters.

Key takeaways

  • SLR has one predictor and one continuous response.
  • The error term separates the fitted conditional mean from an individual observed outcome.
  • Linearity is about unknown coefficients; a transformed predictor can still yield linear regression.

Building a regression model

The workflow moves from data quality to a decision-ready model: collect primary (such as surveys/interviews) or secondary data; preprocess it (derive comparable features, address missing values, and investigate outliers); split observations into training and validation sets (the lecture suggests about 75% training); inspect training-set statistics and plots; choose a functional form; estimate coefficients, commonly by OLS; diagnose assumptions and form; evaluate on validation data; and deploy only when performance is acceptable. Training performance alone can mislead: overfitting occurs when a model performs well on training data but poorly on validation data. The goal is consistent performance, with low bias and low variance.

Key takeaways

  • Data collection, preprocessing, exploration, form choice, estimation, diagnostics, validation, and deployment are connected steps.
  • Keep validation data separate from model fitting to check generalization.
  • Overfitting is a gap between strong training results and poor validation results.

6.2.2 Population Relationship, Estimates, and Prediction

For customers with the same income xx, monthly spending varies. The model describes the mean of that conditional distribution:

E(Y∣X=x)=β0+β1xE(Y\mid X=x)=\beta_0+\beta_1x

The population parameters are unknown, so sample data estimate them as b0b_0 and b1b_1. The estimated regression equation is

y^=b0+b1x\hat y=b_0+b_1x

Here b0b_0 is the estimated intercept and b1b_1 the estimated slope. For a given xx, y^\hat y estimates the mean response and can also serve as a prediction for an individual response. Predictions should remain within the observed predictor range; the fitted pattern alone does not justify extrapolation beyond it.

For Hanumantha’s Canara Bank sample, xˉ=32.89\bar x=32.89 lakhs, yˉ=3852\bar y=3852, b1=85.27b_1=85.27, and b0=1047.37b_0=1047.37. Thus:

y^=1047.37+85.27x\hat y=1047.37+85.27x

Each additional lakh of annual income is associated with an estimated ₹85.27 increase in monthly card spend. At x=35x=35 lakhs:

y^=1047.37+85.27(35)=4032. ⁣ (approximately)\hat y=1047.37+85.27(35)=4032.\!\text{ (approximately)}

This is an estimate based on the fitted sample relationship, not a guarantee for a customer.

Key takeaways

  • The population line uses unknown β\beta parameters; the sample fitted line uses bb estimates.
  • y^\hat y estimates the conditional mean and supplies a point prediction.
  • Interpret slope in the units of yy per one unit of xx; avoid unsupported extrapolation.

6.3.1 Ordinary Least Squares

Ordinary least squares (OLS) chooses the line whose predictions have the smallest sum of squared vertical errors. Squaring prevents positive and negative errors from cancelling and penalizes larger misses more.

SSE⁡(b0,b1)=∑i=1n(yi−y^i)2=∑i=1n[yi−(b0+b1xi)]2\operatorname{SSE}(b_0,b_1)=\sum_{i=1}^{n}(y_i-\hat y_i)^2=\sum_{i=1}^{n}[y_i-(b_0+b_1x_i)]^2

Minimizing this quantity gives:

b1=∑i=1n(xi−xˉ)(yi−yˉ)∑i=1n(xi−xˉ)2,b0=yˉ−b1xˉb_1=\frac{\sum_{i=1}^{n}(x_i-\bar x)(y_i-\bar y)}{\sum_{i=1}^{n}(x_i-\bar x)^2},\qquad b_0=\bar y-b_1\bar x

The calculation follows a clear sequence: compute xˉ,yˉ\bar x,\bar y; for each row compute xi−xˉx_i-\bar x and yi−yˉy_i-\bar y; sum their products for the numerator and sum squared xx deviations for the denominator; calculate b1b_1, then b0b_0. The fitted line passes through (xˉ,yˉ)(\bar x,\bar y).

For the 55-customer income example, the resulting estimates are b1=85.27b_1=85.27 and b0=1047.37b_0=1047.37. A scatterplot with xx on the horizontal axis and yy on the vertical axis is a useful first check for an approximately linear trend. Excel can show a linear trendline, its equation, and R2R^2; its regression tool also produces the broader diagnostics described below.

Key takeaways

  • OLS minimizes squared residuals, not the sum of residuals or absolute residuals.
  • The slope is centered cross-deviation divided by predictor variation.
  • Inspect the data pattern before treating a straight line as an adequate form.

6.3.2 Explained Variation and the Coefficient of Determination

For observation ii, the residual is actual minus fitted response, ei=yi−y^ie_i=y_i-\hat y_i. Three sums of squares partition the response variation for an OLS line with an intercept:

SSE=∑i(yi−y^i)2,SST=∑i(yi−yˉ)2,SSR=∑i(y^i−yˉ)2\mathrm{SSE}=\sum_i(y_i-\hat y_i)^2,\quad \mathrm{SST}=\sum_i(y_i-\bar y)^2,\quad \mathrm{SSR}=\sum_i(\hat y_i-\bar y)^2

SST=SSR+SSE\mathrm{SST}=\mathrm{SSR}+\mathrm{SSE}

SST is total variation around the sample mean; SSR is variation explained by fitted values; SSE is leftover variation around the line. A perfect fit has SSE=0\mathrm{SSE}=0; a line explaining none of the total variation has SSR=0\mathrm{SSR}=0.

The coefficient of determination is

R2=SSRSST=1−SSESSTR^2=\frac{\mathrm{SSR}}{\mathrm{SST}}=1-\frac{\mathrm{SSE}}{\mathrm{SST}}

In this setting it ranges from 00 to 11. It is the proportion of sample variation in yy accounted for by the fitted line, not the proportion of individual outcomes predicted exactly, and not evidence that xx causes yy.

For Hanumantha’s sample, the observation with income ₹22 lakhs has actual spend ₹2,429 and fitted spend ₹2,923, so ei=2429−2923=−494e_i=2429-2923=-494. Across all observations, SSE=40,017,408\mathrm{SSE}=40{,}017{,}408, SSR=74,180,516\mathrm{SSR}=74{,}180{,}516, and SST=114,197,923\mathrm{SST}=114{,}197{,}923 (rounding accounts for a one-unit difference in the sum). Thus R2≈0.650R^2\approx0.650: about 65% of the observed sample variation in monthly spend is explained by the fitted income line.

Key takeaways

  • Residuals are vertical differences between observed and fitted yy values.
  • Total variation divides into explained and unexplained sums of squares.
  • R2R^2 describes in-sample fit; it does not establish causation or guarantee predictive accuracy on new data.

6.3.3 Correlation and R2R^2

In simple linear regression with an intercept, the sample correlation between xx and yy is the signed square root of R2R^2:

rxy=sign⁡(b1)R2r_{xy}=\operatorname{sign}(b_1)\sqrt{R^2}

The sign comes from the slope because squaring removes direction. For Hanumantha’s R2=0.65R^2=0.65 and positive slope, rxy=+0.65≈0.806r_{xy}=+\sqrt{0.65}\approx0.806. Correlation ranges from −1-1 to +1+1: the endpoints indicate perfect negative or positive linear association, while values near zero indicate little linear association. Correlation summarizes direction and strength of linear association; R2R^2 summarizes the share of response variation explained by a fitted model. Unlike the two-variable correlation coefficient, R2R^2 also applies in nonlinear and multiple-predictor settings. Neither proves a causal mechanism.

Key takeaways

  • In SLR with an intercept, ∣r∣=R2|r|=\sqrt{R^2} and the slope gives the sign.
  • Correlation and fit are descriptive; they do not, by themselves, identify cause.
  • Distinguish a plausible explanatory mechanism from an unexplained association.

6.3.4 Error-Term Assumptions

Significance tests rely on assumptions about the population error ε\varepsilon:

Assumption Meaning
Zero conditional mean E(ε∣X=x)=0E(\varepsilon\mid X=x)=0, so E(Y∣X=x)=β0+β1xE(Y\mid X=x)=\beta_0+\beta_1x
Constant variance (homoscedasticity) Var⁡(ε∣X=x)=σ2\operatorname{Var}(\varepsilon\mid X=x)=\sigma^2 for every xx
Independence Errors for different observations are independent
Normality Errors are normally distributed at each xx

Together, the errors have mean zero and common variance σ2\sigma^2; because YY is a linear function of the error, conditional responses are normally distributed around a mean that changes with xx and with the same variance. If β1=0\beta_1=0, the conditional mean does not vary linearly with xx; if β1≠0\beta_1\ne0, a linear mean relationship exists.

Key takeaways

  • The assumptions underpin the usual tt and FF inference for regression.
  • Constant variance means the vertical spread around the line does not systematically change with xx.
  • A significant slope is a statement about a linear association in the population under the model assumptions.

6.3.5 Testing Whether the Slope Is Significant

The central test is H0:β1=0H_0:\beta_1=0 against HA:β1≠0H_A:\beta_1\ne0. Under the null, the mean response does not change linearly with xx. A small pp-value is evidence against that null, conditional on the model assumptions.

Estimate error variance using

MSE=s2=SSEn−2,s=MSE\mathrm{MSE}=s^2=\frac{\mathrm{SSE}}{n-2},\qquad s=\sqrt{\mathrm{MSE}}

There are n−2n-2 error degrees of freedom because the intercept and slope were estimated. The slope standard error and tt statistic under the null are

SE⁡(b1)=s∑i(xi−xˉ)2,t=b1−0SE⁡(b1)\operatorname{SE}(b_1)=\frac{s}{\sqrt{\sum_i(x_i-\bar x)^2}},\qquad t=\frac{b_1-0}{\operatorname{SE}(b_1)}

Use a two-sided tt distribution with n−2n-2 degrees of freedom to obtain the pp-value. For the 55-customer example, SSE=40,017,408\mathrm{SSE}=40{,}017{,}408, so MSE=40,017,408/53≈755,045\mathrm{MSE}=40{,}017{,}408/53\approx755{,}045 and s≈868.9s\approx868.9. With ∑(xi−xˉ)2=10,201\sum(x_i-\bar x)^2=10{,}201, SE⁡(b1)≈868.9/10,201=8.603\operatorname{SE}(b_1)\approx868.9/\sqrt{10{,}201}=8.603. Then t=85.27/8.603≈9.91t=85.27/8.603\approx9.91 (53 degrees of freedom; reported p≈1.1×10−13p\approx1.1\times10^{-13}).

The equivalent SLR overall FF test compares explained and residual mean squares:

MSR=SSR1,F=MSRMSE\mathrm{MSR}=\frac{\mathrm{SSR}}{1},\qquad F=\frac{\mathrm{MSR}}{\mathrm{MSE}}

Its degrees of freedom are 11 and n−2n-2. In simple regression, F=t2F=t^2 for the slope test, so both tests give the same significance conclusion. Significance does not mean the relationship is large, useful, causal, or correctly specified.

Key takeaways

  • Test the population slope, not whether the fitted sample slope happens to be zero.
  • MSE\mathrm{MSE} estimates error variance; the slope test uses its square root in the standard error.
  • In SLR, the one-slope tt test and overall FF test are equivalent.

6.3.6 Reading Regression Summary Output

An analysis tool’s summary brings the preceding calculations together:

Output Interpretation
R2R^2 Fraction of sample response variation explained by the fitted line
Standard error / ss MSE\sqrt{\mathrm{MSE}}, estimated residual standard deviation
ANOVA: SSR, SSE Explained and residual sums of squares
MSR, MSE, FF, Significance FF Overall slope test; df 11 and n−2n-2
Coefficients b0b_0 (intercept) and b1b_1 (slope)
Slope standard error, tt, pp Test of H0:β1=0H_0:\beta_1=0

For the 55-customer example, the tool reports R2=0.65R^2=0.65, s=868.9s=868.9, F=98.25F=98.25, b0=1047.37b_0=1047.37, and b1=85.27b_1=85.27 with slope standard error about 8.68.6 and t=9.91t=9.91. The tool is useful because it automates arithmetic; understanding each quantity makes its output interpretable.

Key takeaways

  • Check that selected YY and XX columns correspond row by row.
  • ANOVA’s residual sum of squares is SSE; “Significance F” is the overall test pp-value.
  • Coefficient output and ANOVA should tell consistent stories about the slope.

6.3.7 Residual Analysis

Residuals are observable clues about errors that cannot be directly observed. Check plots of residuals against xx, fitted values y^\hat y, and standardized residuals. A residual plot should look roughly like a horizontal band around zero if the line and constant-variance assumption are reasonable.

Pattern Possible implication
Roughly horizontal band No obvious nonlinearity or changing spread
Funnel, widening with xx Nonconstant error variance; homoscedasticity may fail
Curved / U-shaped pattern Straight-line mean form may be inadequate; consider a transformed or richer model

For simple regression, residuals against xx and fitted y^\hat y show the same pattern because fitted values are a linear function of xx. The fitted-value plot becomes especially useful with multiple predictors. Standardized residuals are residuals divided by their estimated standard deviation. If normality is plausible, roughly 95% should lie between −2-2 and +2+2; this is a diagnostic clue, not a proof of normality.

Key takeaways

  • Look for structure, changing spread, and unusually large standardized residuals.
  • A visually random band supports—but cannot prove—the assumptions.
  • Residual diagnostics qualify the significance conclusions drawn from model output.

6.3.8 Outliers and Influential Observations

An outlier departs substantially from the trend of the remaining observations. A large absolute standardized residual flags a candidate; the lecture uses values outside about ±2\pm2 as a warning to inspect. Do not automatically delete it. Check for recording or measurement error and correct errors; if valid, it may be a chance extreme or reveal that the model is inadequate.

An influential observation can materially change the fitted line, especially its slope, if removed. It may combine an unusual yy with an extreme xx, or be far out in xx even if close to the trend. Inspect the data source and validity. If valid, examine its context and, where possible, collect additional observations at intermediate xx values to clarify the relationship.

Key takeaways

  • An outlier concerns unusual response relative to the pattern; influence concerns impact on the fitted model.
  • Verify observations before correcting, retaining, or changing the model.
  • Extreme predictor values can be influential even when their residual is not especially large.

6.3.9 Putting Inference and Diagnostics Together

The workflow connects fitting, fit assessment, inference, and assumption checking:

For Hanumantha, the fitted equation is y^=1047.37+85.27x\hat y=1047.37+85.27x, R2=0.65R^2=0.65, correlation is about 0.8060.806, and s=868.9s=868.9. The slope tests are highly significant (t=9.91t=9.91, F=98.25F=98.25), while residual plots show no clear funnel or curved pattern; most standardized residuals lie within ±2\pm2, with two slightly beyond. Statistical evidence, fit, and diagnostics answer different questions and should be read together.

Key takeaways

  • A significant slope does not replace checking fit and assumptions.
  • Use R2R^2 for explained sample variation, tests for evidence of a slope, and residuals for model diagnostics.
  • Draw population conclusions only with the model assumptions and sampling context in view.

6.4.1 Application: Household Size and Monthly Card Spend

The first Excel application uses household size to predict Hanumantha’s customers’ monthly card spend. The scatterplot suggests a positive but noisy linear trend. The fitted equation is y^=2083.7+520.13x\hat y=2083.7+520.13x, with R2=0.396R^2=0.396 (about 0.39); household size explains about 39% of sample variation. Correlation is positive, r≈0.396≈0.63r\approx\sqrt{0.396}\approx0.63. The standard error is 1140.711140.71, so MSE≈1140.712≈1,301,218\mathrm{MSE}\approx1140.71^2\approx1{,}301{,}218. ANOVA gives SSR=45,233,346\mathrm{SSR}=45{,}233{,}346, SSE=68,964,576\mathrm{SSE}=68{,}964{,}576, error df 5353, and F=34.7F=34.7; the pp-value is close to zero. The slope standard error is 88.288.2, so t=520.1/88.2≈5.89t=520.1/88.2\approx5.89, also with a very small pp-value.

Three standardized residuals exceed +2+2 and sit above the line. Residuals against household size show neither a clear funnel nor curvature, but residuals against predicted spend show a trend, suggesting other influences such as annual income. Household size is statistically related to spend, but income explains variation better in its separate model.

Key takeaways

  • A visible positive trend can coexist with substantial scatter and modest R2R^2.
  • A single predictor rarely captures every determinant of spending.
  • Interpret model usefulness separately from whether a slope is statistically significant.

6.4.2 Application: Monthly Spend and Annual Income

For 55 customers, the descriptive and OLS calculations give xˉ=32.89\bar x=32.89 lakhs, yˉ=3852\bar y=3852, and y^=1047.37+85.27x\hat y=1047.37+85.27x. Income and spend have a positive linear association. For annual income ₹35 lakhs, predicted monthly spend is approximately ₹4,032. The fitted equation is intended for the income range represented in the data.

The fit and inferential output are R2=0.65R^2=0.65, correlation 0.8060.806, s=868.9s=868.9, SSE=40,017,408\mathrm{SSE}=40{,}017{,}408, SSR=74,180,516\mathrm{SSR}=74{,}180{,}516, SST=114,197,923\mathrm{SST}=114{,}197{,}923, F=98.25F=98.25, and slope t=9.91t=9.91 with pp close to zero. Residual plots broadly resemble a band; most standardized residuals are within ±2\pm2, with two slightly outside. The relationship is statistically significant and explains a meaningful share, but not all, of variation.

Key takeaways

  • Read coefficient size in business units: ₹85.27 more monthly spend per ₹1 lakh of annual income.
  • The test supports a nonzero population slope under the model assumptions.
  • The remaining 35% of variation and unusual residuals remain relevant to decisions.

Additional Worked Example: Season Pass Holders and PIN-Code Population

Arun relates the number of season pass holders (yy) to PIN-code population (xx). The estimated line is y^=−15.24+0.009x\hat y=-15.24+0.009x. Thus an additional 1,000 residents corresponds to about nine additional season pass holders. ANOVA gives SSR=158,752\mathrm{SSR}=158{,}752, SSE=161,238\mathrm{SSE}=161{,}238, n=30n=30, and error degrees of freedom 2828:

MSR=158752/1=158752,MSE=161238/28≈5759,F=158752/5759≈27.5\mathrm{MSR}=158752/1=158752,\quad \mathrm{MSE}=161238/28\approx5759,\quad F=158752/5759\approx27.5

The FF-test pp-value is close to zero, so reject H0:β1=0H_0:\beta_1=0. The slope standard error is 0.00180.0018, giving t=0.009/0.0018=5.0t=0.009/0.0018=5.0 (the transcript reports 5.255.25, reflecting rounded displayed inputs); the pp-value is very small and leads to the same conclusion. One standardized residual is 3.63.6 while the others are within ±2\pm2, flagging a PIN code well above the line. Apart from that observation, residuals against population look random. Population is related to season-pass counts, but demographics (families and children) and income may also matter. The relationship can help target marketing; it does not alone explain local differences.

Key takeaways

  • The fitted slope translates population differences into expected pass-holder differences.
  • Both the overall FF and slope tt tests indicate a significant linear association.
  • Investigate the high residual and collect other plausible predictors before over-interpreting the line.

Additional Worked Example: Bus Age and Annual Maintenance Cost

For 30 BMTC buses aged 1–5 years, mean age is 3 years (median 2.5); mean annual maintenance cost is ₹6.1 lakhs with standard deviation about ₹2 lakhs and observed costs up to ₹9.5 lakhs. The scatterplot suggests a linear increase. The estimated regression line is

y^=2.2+1.32x\hat y=2.2+1.32x

where yy is annual maintenance cost in lakhs and xx is age in years. Each additional year is associated with about ₹1.32 lakhs higher annual maintenance cost. Though the intercept is 2.2, it must not be read as the cost of a new bus: age zero is outside the observed range, so that is extrapolation. The line is for ages 1 through 5.

The fit is R2=0.87R^2=0.87, hence positive correlation r=0.87≈0.93r=\sqrt{0.87}\approx0.93. Standard error s=0.69s=0.69, so MSE≈0.692=0.48\mathrm{MSE}\approx0.69^2=0.48. ANOVA values are SSR=93.5\mathrm{SSR}=93.5, SSE=13.65\mathrm{SSE}=13.65, SST=107.1\mathrm{SST}=107.1, with regression df 11 and error df 2828:

MSR=93.51,MSE=13.65/28≈0.48,F=93.51/0.48≈191.7\mathrm{MSR}=93.51,\quad \mathrm{MSE}=13.65/28\approx0.48,\quad F=93.51/0.48\approx191.7

The FF-test pp-value is very small. The slope standard error is 0.0950.095, giving t=1.32/0.095≈13.85t=1.32/0.095\approx13.85 and a very small pp-value; reject H0:β1=0H_0:\beta_1=0. All standardized residuals are between −2-2 and +2+2, and the residual plot against age looks random. The line can help quantify maintenance costs for fleet budgeting and replacement planning within the sampled age range.

Key takeaways

  • Strong fit and significant slope support an age–maintenance association in this sample.
  • A plausible coefficient interpretation does not make extrapolation valid.
  • Residual checks and the observed predictor range bound how the model should be used.

6.4.2 Application: Computer Price and Customer Satisfaction

In a 36-response computer satisfaction survey, price (in thousands of rupees) is the predictor and score the response. The trendline is

y^=48.44+0.132x,R2=0.36,r≈+0.59\hat y=48.44+0.132x,\qquad R^2=0.36,\qquad r\approx+0.59

Price explains 36% of score variation. The residual standard error is 5.195.19. The output reports SSR=510.8\mathrm{SSR}=510.8, SSE=914.8\mathrm{SSE}=914.8 in one utterance but then uses 940.8940.8 in the calculation; 940.8/34≈27.7940.8/34\approx27.7, while the reported MSE=26.9\mathrm{MSE}=26.9 corresponds to about 914.8914.8. The reported F=18.9F=18.9 and slope standard error 0.030.03 give t=0.132/0.03=4.4t=0.132/0.03=4.4 (reported 4.364.36); both tests have very small pp-values. The positive slope means a ₹1,000 increase in price is associated with a 0.1320.132 point higher score in these data, even though intuition suggested the opposite. Regression describes the observed sample pattern; unexpected direction merits investigation.

Two standardized residuals are below −2-2 (about −2.48-2.48 and −2.12-2.12), where the model overpredicts scores. Residual plots show no clear funnel or curvature against price, but a trend against score suggests other factors influence satisfaction.

Key takeaways

  • A statistically significant direction can conflict with prior expectations; the fitted model follows the data.
  • R2=0.36R^2=0.36 leaves most score variation unexplained by price alone.
  • The spoken ANOVA sums of squares are inconsistent; the notes preserve that ambiguity.

6.4.3 Application: Household Debt and Income and Monthly Payment

The household survey has about 500 Bengaluru observations. The analysis repeatedly keeps one predictor at a time, so each result is a separate SLR rather than a controlled “effect” holding other variables constant.

Debt and first income

With household debt as yy and first income as xx, the trendline is y^=32976+0.04x\hat y=32976+0.04x and R2=0.31R^2=0.31, or correlation about +0.56+0.56. The standard error is ₹28,619. ANOVA reports SSR=1.86×1011\mathrm{SSR}=1.86\times10^{11} and SSE=4.08×1011\mathrm{SSE}=4.08\times10^{11}, with n−2=498n-2=498 error degrees of freedom; F≈228F\approx228 and pp is close to zero. The coefficient output gives slope about 0.040.04 (one spoken calculation says 0.40.4, inconsistent with the equation and reported t=15t=15; the equation’s 0.040.04 is the coherent displayed estimate). The association is significant but income alone explains only 31% of debt variation. Residuals include values beyond ±2\pm2 and an oval-like pattern, leaving scope for other predictors.

Debt and monthly payment

With debt as yy and monthly payment as xx, the trendline is y^=28792+2.94x\hat y=28792+2.94x, with R2=0.37R^2=0.37 and positive correlation about 0.600.60. Standard error is ₹27,518. ANOVA reports SSR=2.18×1011\mathrm{SSR}=2.18\times10^{11} and SSE=3.8×1011\mathrm{SSE}=3.8\times10^{11}; with 498 error degrees of freedom, MSE=3.8×1011/498≈7.63×108\mathrm{MSE}=3.8\times10^{11}/498\approx7.63\times10^8 and F≈288F\approx288. The slope standard error is 0.1740.174, so t=2.94/0.174≈16.95t=2.94/0.174\approx16.95 and the pp-value is close to zero. Several standardized residuals exceed ±2\pm2 and plots suggest remaining structure. Monthly payment alone explains 37% of debt variation. Since debt can depend on several factors, the slope is an unadjusted association; multiple regression is needed to study several predictors together.

Key takeaways

  • Statistical significance can coexist with modest explanatory power and residual concerns.
  • One-predictor slopes can reflect omitted factors and should not be called isolated causal effects.
  • Unexpected signs, outliers, and residual patterns are prompts to investigate and improve the model.
Revise before your examAll 1 exam tip from this module in one list
  1. A fitted line or high training R2R^2 is not enough for deployment. Check assumptions and validation performance.

Test yourself on this module

Questions from these notes, one at a time, with feedback. Your weak topics are saved.

Set it up free Takes about a minute. Free plans work.