DBB3102 BUSINESS ANALYTICS JAN FEB 2026

Description

SESSION	JAN – FEB 2026
PROGRAM	BACHELOR OF BUSINESS ADMINISTRATION (BBA)
SEMESTER	V
COURSE CODE & NAME	DBB3102  BUSINESS ANALYTICS

 

Assignment Set – 1

 

Q.1. Describe how Business Analytics can help an organization improve its operational efficiency. (10 Marks)

Ans 1.

Business Analytics is the iterative, systematic and statistical exploration of business data to aid in business decision-making. It uses descriptive, predictive and prescriptive analytics that allow organizations to gain insights into what has happened, what might happen and what should happen in order to improve current operations for optimum efficiency.

Descriptive Analytics for Process Understanding

Descriptive analytics gives organizations an understanding of their operational history through dashboards, reports and visual analytics by examining historical data. In manufacturing, descriptive analytics is used to determine which production lines are most likely to produce defects, which shifts are most productive, and most common bottlenecks. Retailers monitor inventory turnover, stock-outs and warehouse capacity to detect inefficiencies. This factual knowledge provides the starting point for improvement.

Predictive Analytics for Proactive Management

Predictive analytics applies statistical models or machine learning to predict the conditions of future operational scenarios. Supply chain managers forecast demand variability based on historical sales data and other factors such as seasonality, economic cycles and promotional schedules, allowing them to position inventory proactively, avoiding stockouts and overstocks. Predictive maintenance models process sensor readings from equipment to monitor early signs of equipment failure, allowing planned maintenance to avoid equipment breakdowns and eliminate production losses due to unplanned downtime.

Prescriptive Analytics for Decision Optimization

Prescriptive analytics takes predictive analytics to the next level by identifying the best decision among all possible options. Vehicle routing solutions optimize fuel use and delivery speed by considering traffic, vehicle loads and time windows. Staff scheduling systems allocate resources to shifts based on forecasted demand, leading to lower costs without compromising service levels. Capital allocation analytics determine the best allocation of investment across multiple efficiency improvement initiatives, yielding the greatest return on efficiency investment.

Real-World Impact and Benefits

Business analytics adoption results in demonstrable improvements in cycle time, operating cost and resource efficiency, as well as improved customer service. Amazon employs analytics to improve each aspect of its distribution operations from order picking paths to choice of carrier. Indian firms, such as Infosys and Mahindra, use analytics to find opportunities for cost savings and increase service quality. Business analytics converts operational management from firefighting to fact-based improvement that is compounding in its efficiency effects. Companies that consistently use descriptive, predictive and prescriptive analytics build a compounding efficiency advantage hard to match by competitors using intuition-based management. Data-based management delivers superior returns that support the investment in analytics across all sizes, industries and business life cycles around the world.

 

Q.2. How can data updating ensure the accuracy and reliability of data? (10 Marks)

Ans 2.

Data updating is the process of updating, adjusting and refreshing data to ensure it remains as current as possible. In business analytics, the validity and usefulness of any analysis is only as good as the data it’s based on, which means effective data updating is essential for reliable business intelligence.

Real-Time and Scheduled Data Updating

Real-time data updating means that systems are always up-to-date with the current state of the business by capturing transactions and events in real time. Retail systems that update stock levels at the point of sale remove the lag in stock levels that occurs with delayed updating. Customer relationship management systems that record and update every customer interaction in real time guarantee that sales personnel have up-to-date information about each customer, avoiding the risk of inappropriate or redundant outreach due to outdated customer data. Data batch updating, which groups and applies data updates at scheduled times, balances the efficiency of data updating against the timeliness necessary for applications where a small delay in the data is not problematic.

Data Validation and Quality Controls

Data updating processes must include validation rules to reject or warn about data that does not meet quality criteria. Numerical range checks ensure that values lie within the anticipated range, protecting analytical data sets from data entry errors. Referential integrity constraints maintain consistent relationships between data tables during updates, avoiding “orphan” records that skew analytical findings. De-duplication rules detect and consolidate duplicate records during updates, preserving the unique identity of entities that is essential for accurate analytics. Logging who changed what data and when creates accountability and allows tracing of data quality problems back to their origin.

Master Data Management and Synchronization

Master data management provides a single source of truth for critical business entities like customers, products, suppliers and locations, so that all systems access the same timely and accurate data. When a company changes a customer’s contact information, master data management will automatically update all systems to reflect the change, eliminating the problem of different systems referring to the same customer with different contact information. This consistency ensures that analytical models work with a consistent view of the business.

Benefits of Systematic Data Updating

Companies that follow systematic data updating processes enjoy far greater analytical accuracy, as they’re operating with the latest information rather than stale data. Banks update credit risk models with the most recent transaction and behavior data to ensure loan decisions are based on up-to-date customer credit information. Rigorous data updating disciplines allow firms to create the high-quality, timely data infrastructure needed for business analytics to produce consistently reliable and actionable insights, so data updating is one of the most critical operational investments for organizations with business analytics.

 

Q.3. Discuss how mobile and location-based Market Basket Analysis could be used in the future. (10 Marks)

Ans 3.

Market Basket Analysis is a data mining approach that uncovers patterns of product associations based on their co-occurrence in the purchase basket, providing insights that retailers can use to improve product positioning, promotional bundling and cross-selling opportunities. Location-based and mobile extensions of market basket analysis greatly broaden the analytical power and potential application of this technique.

Real-Time Basket Analysis with Mobile Shopping

With more and more shopping occurring through mobile apps, retailers can perform market basket analysis at the level of the individual shopping session, rather than after the fact for completed transactions. Mobile applications track user interactions, including the order in which items were viewed, added and removed from the shopping cart and purchased, offering more detailed behavioral insights than point-of-sale data. This session-level data allows real-time recommendations of related products during the same shopping session using basket analysis algorithms that take into account the products viewed and purchased so far in the session as well as those purchased in millions of previous sessions similar to the current one.

Location-Based Market Basket Analysis

Geofencing and GPS location data allows location-sensitive market basket analysis that links purchase data to geographic context. Stores can determine if customers who walk through a certain store aisle end up buying certain product combinations, facilitating physical store design and aisle placement based on customer walking and purchase sequence patterns. Geolocation data shows that customers who visit certain parts of a city are more likely to purchase particular product combinations, allowing hyperlocal store product assortment customization in local stores for the different purchase patterns of each location’s customers.

Cross-Channel and Cross-Retailer Basket Intelligence

Upcoming mobile basket analysis will increasingly be cross-channel and across multiple retailers due to data sharing and anonymized data aggregations. A consumer who shops for ingredients to prepare a particular type of cuisine at the grocery store could get restaurant recommendations or cooking equipment suggestions from partner apps that add to the value proposition of basket intelligence. Basket analysis with loyalty program data from partner merchants can uncover inter-merchant shopping trends that can’t be derived from an individual merchant’s data alone.

Privacy Considerations and Future Potential

The growth of mobile and location-based market basket analysis present privacy challenges that demand clear data handling policies, informed consent and opt-out mechanisms. Companies that treat location and behavioral data ethically and effectively communicate to consumers the benefits of location and mobile analysis will build the trust necessary to support these analytical techniques. The future of market basket analysis is real-time, geospatial and targeted; it offers new opportunities to provide relevant recommendations, enhancing in-store experiences and generating substantial incremental revenue. Consumer privacy compliant mobile basket analysis is one of the most lucrative opportunities in retail analytics. Mobile basket analysis will profoundly impact retail personalisation over the next ten years.

 

Assignment Set – 2

 

Q.4. Discuss in detail how decision trees work in classification problems. (10 Marks)

Ans 4.

A decision tree is a supervised machine learning algorithm that learns classification or regression rules by partitioning the data into subsets based on the values of features. In classification tasks, the objective is to learn rules from the data that can be used to predict the class of new unseen data cases, with decision trees being one of the simplest, most interpretable and popular machine learning algorithms used in business analytics.

Structure of a Decision Tree

A decision tree is a structure with internal nodes, branches and leaf nodes. Internal nodes are tests on an attribute. Branches are the possible outcomes of the test and leaf nodes are the labels assigned to the instances that reach that node after walking the path from the root of the tree. For example, a tree for predicting customer churn could start with a test on the decline in monthly usage, with branches representing high and low monthly declines, and each branch further split on contract length, number of calls to customer service, and payment delinquency until the leaf nodes assign a customer to either the churn or retain class.

Splitting Criteria

At each node, the algorithm chooses the feature and threshold to split the data into the purest subsets containing mostly instances of a single class. Gini impurity is the probability that a randomly chosen instance from the node will be misclassified if it is labeled as the class that it most likely belongs to. Information gain, derived from information theory’s entropy, is the decrease in uncertainty in class labeling upon splitting on a feature.

Pruning and Overfitting Prevention

Unpruned decision trees are prone to overfit due to growing too deeply to perfectly classify all training data, including noise. Pruning overcomes overfitting by removing tree branches that offer little improvement in classification. Pre-pruning prevents further splits if they do not improve performance by at least a threshold amount. Post-pruning builds the entire tree and then eliminates subtrees that fail to enhance generalization performance on validation data. Minimum leaf size constraints avoid very small leaf nodes that represent noise.

Advantages and Business Applications

Decision trees are very transparent, generating simple rules that business users non-technical in machine learning can easily understand and audit. Decision trees are used by banks for credit decisions with regulatory requirements for explainable model operations. Marketers use decision trees to segment customers for marketing. Decision trees are used by medical institutions to predict patient risk for timely treatment. Random forests and gradient boosting build on simple decision trees by training multiple trees in a forest that work together in an ensemble to deliver better accuracy at the expense of some interpretability, for a range of classification tasks in business analytics.

 

Q.5. Explain the Data Mining Process. (10 Marks)

Ans 5.

Data mining is an analytical process designed to explore large data sets in order to discover previously unknown and potentially useful information such as patterns, associations, anomalies, and trends using a variety of mathematical, statistical and machine learning methods. Data mining is structured in a way that ensures the insights and knowledge gained from the data are sound, useful and reflect the business’s goals rather than reflecting anomalies in the data or methodology used to produce the model.

Step One – Business Understanding

The first step in the data mining process is to establish the business opportunity or problem that the data mining will address. This is the point where business objectives are translated to questions for analysis: What drives churn? Which transactions are most likely fraudulent? Which product baskets have highest value? Business knowledge guides the choice of mining method, success criteria and focuses analytical effort on discovering interesting patterns that are relevant to the business, instead of interesting but irrelevant patterns.

Step Two – Data Understanding and Preparation

Data understanding includes gathering available data, profiling data features using descriptive statistics and visualizations, and evaluating data quality in terms of completeness, consistency and accuracy. Data preparation, which accounts for 60-80 percent of project time, involves converting raw data into the “clean” analytic data set used as input to the mining algorithms. This involves dealing with missing data, eliminating duplicate rows, transforming categorical features, normalising numerical features, and generating new features that reflect patterns in data that are relevant to business but not explicitly represented in the raw data.

Step Three – Modeling

Several data mining algorithms are tested on the prepared data and their results compared on the validation data. The choice of algorithm is based on the type of problem (classification, clustering, association rule mining, regression), the type of data and the need for interpretability. Cross-validation methods examine model performance using data not used for training and give a good indication of how the model will perform on new unseen data in a real deployment scenario.

Steps Four and Five – Evaluation and Deployment

Model evaluation determines if the mining process had actually answered the business question and achieved the success criteria. Patterns are presented to business stakeholders for actionability. Deployment embeds models that have been verified to be useful in the business processes where they generate value through automated scoring, alerting or recommendations. Organizations adopting the structured CRISP-DM approach continually deliver superior mining outcomes that deliver tangible business value, reinforce trust in the business to make data-driven decisions, develop the business’s analytical capability to support ongoing competitive assessment and improvement, and show business stakeholders the tangible return on their investment in data mining programs. The CRISP-DM process formalises these steps in a structured iterative process where feedback from each phase feeds into improvements in earlier phases to ensure the final models deployed in production actually align to business needs, deliver continuous value as part of an operational process and benefit from ongoing monitoring and refinement.

 

Q.6. Discuss some of the challenges that organizations may face in managing data. (10 Marks)

Ans 6.

In the digital age, data management is one of the most multifaceted and strategic issues for organisations. With increasing volumes, diversity, and velocity of data and increasingly complex analytical tools, the challenges of managing data assets are growing in terms of technical, organizational and regulatory complexity.

Data Volume, Velocity, and Variety

The rapid increase in data from online transactions, social media, IoT devices, and mobile devices presents storage, processing and management challenges. Relational databases are not well-suited to the scale and diversity of today’s data, and require big data infrastructure such as distributed storage, cloud computing, and parallel processing systems. The speed of streaming data from real-time sources also poses problems for batch-oriented systems. Companies need to constantly upgrade their data management infrastructure to ensure data management capacity does not limit analytics and operations.

Data Quality and Consistency

Data quality issues are widely recognised as the biggest impediment to analytics success. Issues with data quality such as missing data, data duplication, inconsistent data formats, stale data and data entry errors distort analytical results and erode confidence in data-based decision making. Ensuring consistent data definitions across various data systems such as Enterprise Resource Planning (ERP), Customer Relationship Management (CRM) and operational databases is especially problematic in businesses that have acquired other businesses and/or built systems independently within different business units. Data governance policies that define ownership, quality and stewardship of critical data assets are important but difficult to implement and maintain in large organizations.

Data Security and Privacy Compliance

An increasing volume of customer and employee data being processed by organizations poses cybersecurity risks and regulatory compliance challenges. Data breaches result in financial losses, regulatory fines and brand-damaging reputational risk. Meeting privacy laws such as GDPR in Europe and India upcoming data protection law, an organisation must implement technical and organisational measures for data minimisation, consent and data subject rights enforcement. Access to data for analytics while ensuring security and privacy requires the development of robust access control and governance oversight.

Organisational and Cultural Challenges

Data management is also constrained by organisational challenges such as the use of “data silos” where information is not routinely shared across departmental lines, and a lack of data literacy among managers responsible for understanding the analytical insights. Creating the data culture that translates investment in data management into value requires ongoing leadership support, training and investment in data literacy and reward systems that encourage data sharing and decision making based on data insights. Organisations that tackle both technical and cultural data management challenges lay the data foundation that supports the delivery of credible business insights from analytics.