Artificial Intelligence in Business ( MID & FINAL )

```json { "ocrText": "1. Let us consider a closed room, where windows and doors are made with glasses. There is an AC (Air conditioner) placed in the top of the room, which works for 'Temperature Controlling System (TMP) for this room. If the desired temperature set for the room is more than the current temperature, TMP adjusts temperature by dissipating cool air. If the desired temperature set for the room is less than the current temperature, TMP adjusts temperature by dissipating hot air. If the desired temperature set for the room equals the current temperature, TMP performs no operation.\na) Show the PEAS description of the task environment for TMP. [2] CLO1\nb) Relate TMP to distinguish its agent type and describe the type in details with figure. [3] CLO1\n\n2. Consider the following scenario, where start node is S and goal node is D.\nValues for h:\nA 5\nB 6\nC 15\nD 0\nX 8\nY 20\nZ 15\na) Demonstrate uniform cost search to find the solution path from S to D. [5] CLO2\nb) Interpret a search tree using A* algorithm to find the solution path from S to D. [5]\n\n3. a) Define Artificial Intelligence (AI) and describe its major branches. Provide real-life examples. [3] CLO1\nb) Differentiate between Classification and Regression based on their working principles, advantages, and applications. [2] CLO1\n\n4. You are designing an AI-powered delivery robot for a logistics company. The robot must navigate city roads, avoid obstacles, and deliver packages to customer locations. During design discussions, your team debates whether the robot's environment should be considered fully observable and deterministic or partially observable and stochastic.\na) Illustrate which type of environment best represents the real-world city delivery scenario. [3] CLO2\nb) Demonstrate your answer by comparing it with a fully observable, deterministic environment such as a chess game or an automated sorting system. [2]\n\n1. A telecom company wants to reduce customer churn. They collected data on monthly usage, contract length, and monthly bill amount. They also classify each customer as “Likely to Churn\" or "Not Likely to Churn.\"\nUsing this business scenario, explain how the company can:\n• Use regression to predict monthly bill amount for new customers\n• Use correlation to identify whether usage or contract length has a stronger relationship with churn behavior\n• why regression and classification serve different business objectives\nYour answer must be business-focused with proper reasoning. [10] CLO1\n\n2. A bank uses an AI model to classify loan applicants as High Risk or Low Risk. After testing, the confusion matrix is:\nPredicted High Risk | Predicted Low Risk\nActual High Risk | 120 | 30\nActual Low Risk | 25 | 225\nCompute the following with full steps:\n1. Accuracy.\n2. Error Rate\n3. True Positive Rate (TPR)\n4. True Negative Rate (TNR)\nFinally, explain in business terms what these metrics tell the bank about the reliability of the loan approval model. [10] CLO2\n\n3. A clothing retailer wants to predict whether a new customer will purchase a Premium or Regular product based on similarity to past customers. Dataset:\nCustomer | Income(k$) | Score | Product\nC1 | 30 | 40 | Regular\nC2 | 80 | 85 | Premium\nC3 | 50 | 60 | Premium\nC4 | 28 | 30 | Regular\nC5 | 75 | 70 | Premium\nNew customer: Income = 60k, Score = 78\nUsing k = 3 (Euclidean distance):\n• Compute distance from each existing customer\n• Identify 3 nearest neighbors\n• Predict whether the new customer will choose Premium or Regular\n• Explain why kNN is useful for business customer segmentation\nAll steps must be shown clearly. [10] CLO3\n\n4. A fashion retail chain wants to increase sales using AI. They collect transaction data from their stores. The transactional dataset is:\n• T1: {T-Shirt, Jeans, Sneakers}\n• T2: {T-Shirt, Jacket}\n• T3: {Jeans, Sneakers}\n• T4: {Jacket, Sneakers}\n• T5: {T-Shirt, Jeans}\nUsing this dataset:\n1. Generate frequent item-sets using Apriori (minimum support = 40%).\n2. Extract two strong association rules (minimum confidence = 60%).\n3. Describe how deep learning creates value in fashion retail through:\no Demand forecasting\no Dynamic pricing\no Personalized style recommendations\no Trend prediction [10] CLO4" } ```