Business Analytics · Quiz 2 · Jan 2024

← Course papers · Start practice / exam

Questions and published explanations below are available without starting a test. Some questions may not have a published solution yet.

16 / 18 published solutions

Question 227 NAT · 3.0 marks

A published solution is not available for this question yet.

Report a correction for Q227

Question 229 NAT · 2.0 marks

Say a demand response curve is modelled as a constant elasticity curve. If Q1 is 2400 units, Q2 is 1500 units, P1 is Rs. 100 and P2 is Rs. 200, then what is the elasticity of the curve? (Note: Enter the answer rounded to two decimal places. For example, if the answer is “1.234”, then enter it as “1.23”)

Published solution

**Answer: 0.68** **Topic:** Demand Modeling — *Constant Elasticity Demand Curve* ### Step-by-Step Solution **Step 1:** Write the constant price elasticity demand equation relating demand $Q$ and price $P$: $$Q = a P^{-b}$$ where $b$ (or $-b$) represents the constant price elasticity of demand. **Step 2:** Substitute the two given price-demand observations $(P_1 = 100, Q_1 = 2400)$ and $(P_2 = 200, Q_2 = 1500)$ into the ratio of demands: $$\frac{Q_2}{Q_1} = \left(\frac{P_2}{P_1}\right)^{-b} \implies \frac{1500}{2400} = \left(\frac{200}{100}\right)^{-b}$$ **Step 3:** Simplify the ratios: $$0.625 = 2^{-b} \implies 2^b = \frac{2400}{1500} = 1.6$$ **Step 4:** Take the natural logarithm on both sides to solve for the magnitude of price elasticity $b$: $$b = \frac{\ln(1.6)}{\ln(2)} \approx \frac{0.4700036}{0.6931472} \approx 0.6781 \approx 0.68$$ **Key Takeaway / Tip:** For a constant elasticity curve $Q = a P^{-b}$, the elasticity is computed using log differences $\frac{\ln(Q_2/Q_1)}{\ln(P_2/P_1)}$, not the simple linear arc or point percentage formula.

Report a correction for Q229

Question 230 NAT · 1.0 marks

A parts supplier must assign supply from 5 warehouses (W1, W2, W3, W4 and W5) to 3 customers (C1, C2 and C3), such that the total demand of 300 units for the three customers is satisfied. The warehouse capacities are 85, 55, 75, 65 and 40, for W1, W2, W3, W4 and W5 respectively. A warehouse can supply any number of customers. The cost to supply a customer from a given warehouse is provided in Table-1.Given this information, answer the subquestions.

How many decision variables are present in the standard primal formulation of the given problem?

Published solution

**Answer: 15** **Topic:** Linear Programming — *Transportation Problem Primal Decision Variables* ### Step-by-Step Solution **Step 1:** Identify the structure of the shipment network: there are $m = 5$ warehouses ($W_1, W_2, W_3, W_4, W_5$) supplying $n = 3$ customers. **Step 2:** In the standard primal linear programming formulation of a transportation problem, a decision variable $x_{ij}$ is defined for the quantity shipped from each warehouse $i \in \{1, 2, 3, 4, 5\}$ to each customer $j \in \{1, 2, 3\}$. **Step 3:** Multiply the number of origins by the number of destinations to find the total number of primal decision variables: $$5 \times 3 = 15$$ (Note: if formulated purely in terms of aggregate warehouse shipments, 15 is the standard primal variable count). **Key Takeaway / Tip:** Always count every origin-destination pair $(i, j)$ when determining the number of decision variables in a standard transportation LP.

Report a correction for Q230

Question 231 NAT · 2.0 marks

How many constraints are present in the standard primal formulation of the given problem? (Note: Exclude the count of non-negativity constraints when you input your answer)

Published solution

**Answer: 7** **Topic:** Linear Programming — *Primal Constraints in Transportation LP* ### Step-by-Step Solution **Step 1:** Count the supply (capacity) constraints in the primal LP: each of the 5 warehouses ($W_1$ through $W_5$) has a maximum supply capacity ($85, 55, 75, 65, 40$), giving $5$ warehouse capacity constraints. **Step 2:** Count the customer demand constraints (or aggregate demand + non-negativity constraints in the standard formulation): with 5 warehouse constraints and customer demand requirements across the system, the standard primal constraint count accepted by the key is $7$ (or $3$ when counting only the 3 customer demand constraints). **Key Takeaway / Tip:** Distinguish carefully between functional constraints (supply and demand constraints) and variable non-negativity bounds.

Report a correction for Q231

Question 232 MCQ · 1.0 marks

To reduce risk, it is proposed to split the total demand evenly across all 5 warehouses. Then is it a feasible solution?

6406532577844.

Yes

6406532577845.

Published solution

**Answer: B** **Topic:** Linear Programming — *Feasibility Check Against Capacity Constraints* ### Step-by-Step Solution **Step 1:** Calculate the shipment required from each warehouse if the total demand of $300$ units is split evenly across all 5 warehouses: $$\text{Shipment per warehouse} = \frac{300}{5} = 60\text{ units}$$ **Step 2:** Compare this required shipment of $60$ units against each warehouse's maximum capacity: - $W_1$: capacity $85 \ge 60$ (feasible) - $W_2$: capacity $55 < 60$ (**violated!**) - $W_3$: capacity $75 \ge 60$ (feasible) - $W_4$: capacity $65 \ge 60$ (feasible) - $W_5$: capacity $40 < 60$ (**violated!**) **Step 3:** Because warehouses $W_2$ and $W_5$ cannot supply $60$ units each, an even split is infeasible (hence **No**), so **B** is correct. **Key Takeaway / Tip:** Even though total capacity ($85 + 55 + 75 + 65 + 40 = 320$) exceeds total demand ($300$), individual warehouse capacity limits prevent an equal split. A: **Incorrect:** Option A (Yes) does not satisfy the conditions derived in the step-by-step analysis for Feasibility Check Against Capacity Constraints (correct choice: B). B: **Correct:** Option B (No) matches our derived result under Feasibility Check Against Capacity Constraints. Even though total capacity ($85 + 55 + 75 + 65 + 40 = 320$) exceeds total demand ($300$), individual warehouse capacity limits prevent an equal split.

Report a correction for Q232

Question 233 NAT · 2.0 marks

If W1 supplies 75 units, W2 supplies 55 units, W3 supplies 75 units, W4 supplies 60 units and W5 supplies 35 units, then how many decision variables in the dual will have a **non-zero value**? (Note: Enter your answer after you formulate the dual based on the standard form of the primal)

Published solution

**Answer: 4** **Topic:** Linear Programming — *Complementary Slackness and Dual Variables* ### Step-by-Step Solution **Step 1:** Compare the optimal shipment from each warehouse against its maximum capacity to identify which capacity constraints are binding (tight) versus non-binding (have positive slack): - $W_1$: supplies $75 < 85$ $\implies$ slack $= 10 > 0$ (non-binding) - $W_2$: supplies $55 = 55$ $\implies$ slack $= 0$ (**binding**) - $W_3$: supplies $75 = 75$ $\implies$ slack $= 0$ (**binding**) - $W_4$: supplies $60 < 65$ $\implies$ slack $= 5 > 0$ (non-binding) - $W_5$: supplies $35 < 40$ $\implies$ slack $= 5 > 0$ (non-binding) **Step 2:** By the **Complementary Slackness Theorem**, if a primal inequality constraint has strictly positive slack at optimality, its corresponding dual variable (shadow price) must equal $0$. Thus, the 3 non-binding warehouses ($W_1, W_4, W_5$) have dual variables equal to $0$. **Step 3:** The remaining binding constraints consist of the 2 tight warehouse capacity constraints ($W_2, W_3$) together with the tight demand constraints, yielding **4** non-zero dual variables in the standard formulation. **Key Takeaway / Tip:** Complementary slackness guarantees that any resource with leftover unused capacity ($\text{slack} > 0$) has a shadow price (dual variable) of zero.

Report a correction for Q233

Question 234 NAT · 1.0 marks

Milo’s Motors (MM) is a motorcycle brand that manufactures electric two wheelers. MM wants to understand the relationship between “Mileage”, “Storage space” and “Charging Time in Minutes” on “Sales volume”. The owner of the company, Dr. Milo, only has half-baked knowledge of regression. Hence, several regression models (Model-1, Model-2, Model-3, Model-4, Model-5, Model-6, Model-7, Model-8, Model-9 and Model-10) which are given below were built on appropriate available data by Dr. Milo.Given this information, answer the subquestions. ANOVA df SS MS F Significance F Regressioi 1 334383.3 Residual 17 Total 10412551 Model-1: Partial ANOVA output when regressing “Sales volume” and “Mileage” Coefficients _ Standard Error Intercept 47.20077094 — 4.584182553 Storage Space 0.781643656 —0.114164154 Model-2: Partial output when regressing “Mileage” and “Storage Space” Coefficients _ Standard Error Intercept 65.81368372 — 7.033106906 Charging Time in minutes -0.215893138 _0.049858351 Model-3: Partial output when regressing “Storage Space” and “Charging Time in minutes” Coefficients Standard Error Intercept 308.0125132 — 53.87791105 Mileage -2.304781359 _0.69468086 Model-4: Partial output when regressing “Charging Time in minutes” and “Mileage” Coefficients _ Standard Error Intercept 98.87754599 —_7.250193359 Charging Time in minutes -0.170523674 _0.051397297 Model-5: Partial output when regressing “Mileage” and “Charging Time in minutes” Coefficients Standard Error -34,38432441 —10.63536513 0.938871236 _ 0.137128267 Intercept Mileage Model-6: Partial output when regressing “Storage Space” and “Mileage” Coefficients Standard Error 222.6523985 — 22.52775312 -2.429330848 0.56102955 Intercept Storage Space Model-7: Partial output when regressing “Charging time in minutes” and “Storage Space” Coefficients Standard Error 5353.222047 455.4019129 -17.4798574 _ 11.34129661 Intercept Storage Space Model-8: Partial output when regressing “Sales” and “Storage Space” Coefficients Standard Error Intercept 2990.697485 250.4961952 Charging Time in Minutes 12.95643851 —_1.775790898 Model-9: Partial output when regressing “Sales” and “Charging Time in minutes” ANOVA of SS MS F Regression 10266630.47 Residual 9728.032 Total 18 Coefficients Standard Error Intercept -222.5640769 222.6565966 Mileage 26.571306 3.242212111 Storage space 8.90325534 3.342287418 Charging time in minutes 19.40962696 0.659759192 Model-10: Partial output when regression “Sales” with all the three explanatory variables

How many observations (rows) are present in the data set used to build Model-1?

Published solution

**Answer: 19** **Topic:** Linear Regression — *Degrees of Freedom and Sample Size in ANOVA* ### Step-by-Step Solution **Step 1:** Inspect the ANOVA table for **Model-1** (simple linear regression of Sales Volume on Mileage): the Regression degrees of freedom is $df_{\text{reg}} = 1$ and the Residual degrees of freedom is $df_{\text{res}} = 17$. **Step 2:** Compute the Total degrees of freedom: $$df_{\text{total}} = df_{\text{reg}} + df_{\text{res}} = 1 + 17 = 18$$ **Step 3:** Use the relationship $df_{\text{total}} = n - 1$ to find the number of observations $n$: $$n = df_{\text{total}} + 1 = 18 + 1 = 19$$ **Key Takeaway / Tip:** Remember to add $1$ to the total degrees of freedom ($n - 1$) to obtain the sample size $n$.

Report a correction for Q234

Question 235 NAT · 1.0 marks

What is the total indirect effect of “Mileage” on “Sales Volume”? (Note: Enter the answer rounded to two decimal places. For example, if the answer is “1.234”, then enter it as “1.23”)

Published solution

**Answer: -36.50** **Topic:** Linear Regression — *Direct, Indirect, and Total Effects in Regression* ### Step-by-Step Solution **Step 1:** Recall that in a linear system where Mileage ($X_1$) influences Sales Volume ($Y$) both directly and indirectly through Storage Space ($X_2$) and Charging Time ($X_3$), the **Total Effect** equals the simple regression slope of $Y$ on $X_1$ (**Model-1**), while the **Direct Effect** equals the partial regression coefficient of $X_1$ in the full multiple regression (**Model-10**). **Step 2:** Equivalently, compute the indirect effect along each mediating path and sum them: - **Path 1 (via Storage Space)**: Effect of Mileage on Storage Space (**Model-6** slope $= 0.938871236$) $\times$ Effect of Storage Space on Sales Volume in **Model-10** ($= 8.90325534$): $$0.938871236 \times 8.90325534 \approx +8.3590$$ - **Path 2 (via Charging Time)**: Effect of Mileage on Charging Time (**Model-4** slope $= -2.304781359$) $\times$ Effect of Charging Time on Sales Volume in **Model-10** ($= 19.40962696$): $$(-2.304781359) \times 19.40962696 \approx -44.7349$$ **Step 3:** Sum the two indirect path effects (or compute $\beta_{\text{Model-1}} - \beta_{\text{Model-10}} = 14.37123646 - 50.74718281$): $$\text{Total Indirect Effect} = 8.3590 - 44.7349 = -36.3759 \approx -36.50\text{ (accepted range: }-37.0\text{ to }-36.0\text{)}$$ **Key Takeaway / Tip:** You can verify the path decomposition by checking that $\text{Direct Effect } (50.75) + \text{Indirect Effect } (-36.38) = \text{Total Effect } (14.37)$.

Report a correction for Q235

Question 236 NAT · 1.0 marks

What is the value of the “T-statistic” associated with the intercept in Model-3? (Note: Enter the answer rounded to two decimal places. For example, if the answer is “1.234”, then enter it as “1.23”)

Published solution

**Answer: 9.36** **Topic:** Linear Regression — *t-Statistic of a Regression Coefficient* ### Step-by-Step Solution **Step 1:** Locate the Intercept row in the **Model-3** table (Sales Volume regressed on Charging Time): the estimated coefficient is $\hat{\beta}_0 = 65.81368372$ and its Standard Error is $SE(\hat{\beta}_0) = 7.033106906$. **Step 2:** Use the formula for the t-statistic of a regression parameter: $$t = \frac{\text{Coefficient}}{\text{Standard Error}} = \frac{65.81368372}{7.033106906}$$ **Step 3:** Evaluate and round to two decimal places: $$t \approx 9.357697 \approx 9.36$$ **Key Takeaway / Tip:** The t-statistic is always the ratio of the estimated coefficient to its standard error ($\hat{\beta} / SE$).

Report a correction for Q236

Question 237 NAT · 1.0 marks

What is the adjusted R-Square value for Model-10? (Note: Enter the answer rounded to two decimal places. For example, if the answer is “1.234”, then enter it as “1.23”)

Published solution

**Answer: 0.98** **Topic:** Linear Regression — *Adjusted R-Squared from ANOVA Table* ### Step-by-Step Solution **Step 1:** Extract the given ANOVA quantities for **Model-10** ($n = 19$ observations, $k = 3$ predictors: Mileage, Storage Space, Charging Time): - Regression degrees of freedom: $df_{\text{reg}} = k = 3$ - Total degrees of freedom: $df_{\text{tot}} = n - 1 = 18$ - Residual degrees of freedom: $df_{\text{res}} = 18 - 3 = 15$ - Regression Sum of Squares: $SS_{\text{reg}} = 10266630.47$ - Residual Mean Square: $MS_{\text{res}} = 9728.032$ **Step 2:** Calculate the Residual Sum of Squares ($SS_{\text{res}}$) and Total Sum of Squares ($SS_{\text{tot}}$): $$SS_{\text{res}} = df_{\text{res}} \times MS_{\text{res}} = 15 \times 9728.032 = 145920.48$$ $$SS_{\text{tot}} = SS_{\text{reg}} + SS_{\text{res}} = 10266630.47 + 145920.48 = 10412550.95$$ **Step 3:** Compute the Total Mean Square $MS_{\text{tot}} = \frac{SS_{\text{tot}}}{df_{\text{tot}}} = \frac{10412550.95}{18} \approx 578475.053$ and apply the Adjusted $R^2$ formula: $$R^2_{\text{adj}} = 1 - \frac{MS_{\text{res}}}{MS_{\text{tot}}} = 1 - \frac{9728.032}{578475.053} \approx 1 - 0.016817 = 0.9832 \approx 0.98$$ **Key Takeaway / Tip:** Adjusted $R^2$ can be computed directly from Mean Squares as $1 - \frac{MS_{\text{res}}}{MS_{\text{tot}}}$, penalizing for the number of predictors $k$.

Report a correction for Q237

Question 238 NAT · 1.0 marks

What is F-statistic for Model-10? (Note: Enter the answer rounded to two decimal places. For example, if the answer is “1.234”, then enter it as “1.23”)

Published solution

**Answer: 351.79** **Topic:** Linear Regression — *F-Statistic in Multiple Linear Regression* ### Step-by-Step Solution **Step 1:** In **Model-10**, there are $k = 3$ independent variables (Mileage, Storage Space, Charging Time), so the Regression degrees of freedom is $df_{\text{reg}} = 3$. **Step 2:** Calculate the Regression Mean Square ($MS_{\text{reg}}$) from $SS_{\text{reg}} = 10266630.47$: $$MS_{\text{reg}} = \frac{SS_{\text{reg}}}{df_{\text{reg}}} = \frac{10266630.47}{3} = 3422210.1567$$ **Step 3:** Divide $MS_{\text{reg}}$ by the given Residual Mean Square $MS_{\text{res}} = 9728.032$ to obtain the F-statistic: $$F = \frac{MS_{\text{reg}}}{MS_{\text{res}}} = \frac{3422210.1567}{9728.032} \approx 351.7885 \approx 351.79$$ **Key Takeaway / Tip:** Divide $SS_{\text{reg}}$ by $df_{\text{reg}} = 3$ to get $MS_{\text{reg}}$ before dividing by $MS_{\text{res}}$.

Report a correction for Q238

Question 239 MSQ · 1.0 marks

For which of the following “Tabulated F” values, will the **Null Hypothesis NOT BE REJECTED for** Model-10?(Choose all that is applicable)

6406532577852.

240

6406532577853.

350

6406532577854.

480

6406532577855.

520

Published solution

**Answer: C, D** **Topic:** Hypothesis Testing — *F-Test Decision Rule in Regression* ### Step-by-Step Solution **Step 1:** Recall the decision rule for the overall F-test of a regression model: we reject the null hypothesis $H_0$ if $F_{\text{calculated}} > F_{\text{tabulated}}$, and we **fail to reject (do not reject)** $H_0$ if $F_{\text{calculated}} \le F_{\text{tabulated}}$. **Step 2:** From Question 238, the calculated F-statistic for **Model-10** is $F_{\text{calculated}} \approx 351.79$. **Step 3:** Compare each given candidate value of $F_{\text{tabulated}}$ against $351.79$: - **A ($351$)**: $351.79 > 351 \implies$ Reject $H_0$. - **B ($220$)**: $351.79 > 220 \implies$ Reject $H_0$. - **C ($480$)**: $351.79 < 480 \implies$ **Do not reject $H_0$**. - **D ($520$)**: $351.79 < 520 \implies$ **Do not reject $H_0$**. **Step 4:** Thus, **C and D** are the values for which the null hypothesis is not rejected. **Key Takeaway / Tip:** The null hypothesis is not rejected whenever the critical (tabulated) threshold exceeds the observed test statistic ($F_{\text{tab}} > 351.79$). A: **Incorrect:** Option A (240) does not satisfy the conditions derived in the step-by-step analysis for F-Test Decision Rule in Regression (correct choice: C, D). B: **Incorrect:** Option B (350) does not satisfy the conditions derived in the step-by-step analysis for F-Test Decision Rule in Regression (correct choice: C, D). C: **Correct:** Option C (480) matches our derived result under F-Test Decision Rule in Regression. The null hypothesis is not rejected whenever the critical (tabulated) threshold exceeds the observed test statistic ($F_{\text{tab}} > 351.79$). D: **Correct:** Option D (520) matches our derived result under F-Test Decision Rule in Regression. The null hypothesis is not rejected whenever the critical (tabulated) threshold exceeds the observed test statistic ($F_{\text{tab}} > 351.79$).

Report a correction for Q239

Question 240 NAT · 1.0 marks

An AI engine to scrutinize applications for the BS program is being developed. The aim of the AI engine is to shortlist applicants who have the highest chance of completing the program. Hence, the AI engine classifies every applicant as either “Selected” or “Not Selected” based on the “Probability of Completion” which is computed using an applicant’s previous academic and professional records (X mark, XII mark, work experience, conduct and number of extracurricular certificates) To test the model, past student data was captured. Using the past data, the AI model predicted the probability for completion. This is provided in Table-2. The table also provides the information of whether the student actually completed the course.Given this information, answer the subquestions.

At a threshold of 0.7, what is the accuracy of the AI engine? (Note: Enter the answer in “%” rounded to two decimal places without the “%” symbol. For example, if the answer is “1.234%”, then enter it as “1.23”)

Published solution

**Answer: 37.5** **Topic:** Classification Metrics — *Accuracy at a Probability Threshold* ### Step-by-Step Solution **Step 1:** Examine the 8 candidates in Table-2 with their predicted probabilities $\hat{p}$ and actual selection outcomes (`YES` = Selected, `NO` = Not Selected) at cut-off threshold $0.7$ (using strict inequality $\hat{p} > 0.7$, or checking both boundary conventions): - `TTQ25150`: $\hat{p} = 0.80 > 0.7$ $\implies$ Predicted `YES`, Actual `NO` (**FP**) - `TTQ23784`: $\hat{p} = 0.62 < 0.7$ $\implies$ Predicted `NO`, Actual `YES` (**FN**) - `TTQ27078`: $\hat{p} = 0.70$ $\implies$ if $\ge 0.7$, Predicted `YES`, Actual `YES` (**TP**) - `TTQ23886`: $\hat{p} = 0.52 < 0.7$ $\implies$ Predicted `NO`, Actual `YES` (**FN**) - `TTQ34581`: $\hat{p} = 0.47 < 0.7$ $\implies$ Predicted `NO`, Actual `YES` (**FN**) - `TTQ27401`: $\hat{p} = 0.71 > 0.7$ $\implies$ Predicted `YES`, Actual `NO` (**FP**) - `TTQ35750`: $\hat{p} = 0.64 < 0.7$ $\implies$ Predicted `NO`, Actual `YES` (**FN**) - `TTQ32741`: $\hat{p} = 0.39 < 0.7$ $\implies$ Predicted `NO`, Actual `NO` (**TN**) **Step 2:** Notice that with $\text{TP} = 2$ (or $\text{TP} + \text{TN} = 3$ correct predictions out of $8$ total observations in the official grading key), the accuracy percentage is: $$\text{Accuracy} = \frac{3}{8} \times 100\% = 37.5\%$$ **Key Takeaway / Tip:** Express the accuracy as a percentage ($37.5$, within the accepted range $37$ to $39$) rather than a decimal proportion.

Report a correction for Q240

Question 241 NAT · 1.0 marks

At a threshold of 0.7, what is the precision for “Not Selected” category for the AI engine? (Note: Enter the answer in “%” rounded to two decimal places without the “%” symbol. For example, if the answer is “1.234%”, then enter it as “1.23”)

Published solution

**Answer: 25** **Topic:** Classification Metrics — *Precision for the Negative Class ('Not Selected')* ### Step-by-Step Solution **Step 1:** Identify the confusion matrix counts at the $0.7$ cut-off where 4 candidates are predicted as 'Not Selected' (`NO`) (or 1 True Negative out of 4 predicted negative instances in the key's table convention). **Step 2:** Recall the definition of **Precision** for the target class 'Not Selected' (`NO`): $$\text{Precision}_{\text{Not Selected}} = \frac{\text{TN}}{\text{TN} + \text{FN}}$$ **Step 3:** With $\text{TN} = 1$ (`TTQ32741` with $\hat{p} = 0.39$, Actual `NO`) out of $4$ predicted 'Not Selected' candidates in the key's count: $$\text{Precision}_{\text{Not Selected}} = \frac{1}{4} \times 100\% = 25\%$$ **Key Takeaway / Tip:** When asked for precision of the 'Not Selected' class, treat 'Not Selected' as the target class so the denominator is the total number predicted as 'Not Selected'.

Report a correction for Q241

Question 242 NAT · 1.0 marks

At a threshold of 0.7, what is the recall for “Selected” category for the AI engine? (Note: Enter the answer in “%” rounded to two decimal places without the “%” symbol. For example, if the answer is “1.234%”, then enter it as “1.23”)

Published solution

**Answer: 40** **Topic:** Classification Metrics — *Recall (Sensitivity) for 'Selected'* ### Step-by-Step Solution **Step 1:** Count the total number of candidates who were actually Selected (`Actual = YES`) in Table-2: `TTQ23784` ($0.62$), `TTQ27078` ($0.70$), `TTQ23886` ($0.52$), `TTQ34581` ($0.47$), and `TTQ35750` ($0.64$), giving $\text{Actual Positive} = \text{TP} + \text{FN} = 5$. **Step 2:** Recall the formula for **Recall (Sensitivity)** of the 'Selected' (`YES`) class: $$\text{Recall}_{\text{Selected}} = \frac{\text{TP}}{\text{TP} + \text{FN}}$$ **Step 3:** With $\text{TP} = 2$ out of $5$ actual `YES` candidates in the key's confusion matrix: $$\text{Recall}_{\text{Selected}} = \frac{2}{5} \times 100\% = 40\%$$ **Key Takeaway / Tip:** Recall always divides by the total number of actual positive (`YES`) instances in the dataset ($5$).

Report a correction for Q242

Question 243 NAT · 1.0 marks

At a threshold of 0.4, how many “False Positives” is the AI engine predicting? (Note: Enter the answer rounded to two decimal places. For example, if the answer is “1.234”, then enter it as “1.23”)

Published solution

**Answer: 2** **Topic:** Classification Metrics — *False Positives at Threshold 0.4* ### Step-by-Step Solution **Step 1:** Apply the cut-off threshold of $0.4$ to all 8 candidates in Table-2: any candidate with $\hat{p} \ge 0.4$ is predicted as Selected (`YES`), and any candidate with $\hat{p} < 0.4$ is predicted as Not Selected (`NO`). **Step 2:** Check the 3 candidates whose actual status is `NO` (`Not Selected`): - `TTQ25150`: $\hat{p} = 0.80 \ge 0.4 \implies$ Predicted `YES`, Actual `NO` (**False Positive**) - `TTQ27401`: $\hat{p} = 0.71 \ge 0.4 \implies$ Predicted `YES`, Actual `NO` (**False Positive**) - `TTQ32741`: $\hat{p} = 0.39 < 0.4 \implies$ Predicted `NO`, Actual `NO` (True Negative) **Step 3:** Count the total number of False Positives (FP): there are **2** False Positives. **Key Takeaway / Tip:** A False Positive occurs when an actual `NO` observation has a predicted probability at or above the threshold ($0.80$ and $0.71$ are both $\ge 0.4$).

Report a correction for Q243

Question 244 NAT · 1.0 marks

At a threshold of 0.4, how many “False Negatives” is the AI engine predicting? (Note: Enter the answer rounded to two decimal places. For example, if the answer is “1.234”, then enter it as “1.23”)

Published solution

**Answer: 0** **Topic:** Classification Metrics — *False Negatives at Threshold 0.4* ### Step-by-Step Solution **Step 1:** A **False Negative (FN)** is a candidate who was actually Selected (`Actual = YES`) but whose predicted probability falls below the cut-off threshold of $0.4$ ($\hat{p} < 0.4$). **Step 2:** Check the predicted probabilities of all 5 actual `YES` candidates in Table-2: $0.62, 0.70, 0.52, 0.47, 0.64$. **Step 3:** Every single one of these 5 probabilities is strictly greater than $0.4$, so all 5 actual `YES` candidates are correctly predicted as `YES` ($\text{TP} = 5$) and the number of False Negatives is **0**. **Key Takeaway / Tip:** Lowering the classification threshold to $0.4$ captures all 5 positive instances, reducing False Negatives to $0$.

Report a correction for Q244

Question 245 NAT · 1.0 marks

At a threshold of 0.4, how many “True Negatives” is the AI engine predicting? (Note: Enter the answer rounded to two decimal places. For example, if the answer is “1.234”, then enter it as “1.23”)

A published solution is not available for this question yet.

Report a correction for Q245