The document describes several case studies completed as part of a business analytics course. The case studies covered topics like social media metrics for a gym, car performance analysis, employee salary prediction, fraud detection, stock price prediction, product recommendations, online marketing campaigns, and demand forecasting for a bicycle rental company. Machine learning techniques like regression, neural networks, support vector machines, and ensemble models were applied to solve problems in various domains like healthcare, retail, and transportation.
Почему издателям нужно думать о мобильных платформах?Vsevolod Pulya
Выступление на саммите Ассоциации независимых региональных издателей в Анапе, март 2013 года.
Приложения, мобильные сайты, адаптивный дизайн: почему издателям нужно думать о мобильных платформах?
Почему издателям нужно думать о мобильных платформах?Vsevolod Pulya
Выступление на саммите Ассоциации независимых региональных издателей в Анапе, март 2013 года.
Приложения, мобильные сайты, адаптивный дизайн: почему издателям нужно думать о мобильных платформах?
Data Science Salon: Adopting Machine Learning to Drive Revenue and Market ShareFormulatedby
The race is on to gain strategic and proprietary insights into changes in customer preferences before your competitors. This workshop will cover how and why machine learning is the tool for marketers to drive revenue and increase market share. The adoption of machine learning does not happen overnight. We will discuss the Five Es of machine learning maturity – Educating, Exploring, Engaging, Executing and Expanding. Hear real-world examples of using machine learning to accelerate revenue, identify new customers and introduce new products based on machine learning capabilities.
Next DSS MIA Event - https://datascience.salon/miami/
Data Science Salon: Adopting Machine Learning to Drive Revenue and Market ShareFormulatedby
The race is on to gain strategic and proprietary insights into changes in customer preferences before your competitors. This workshop will cover how and why machine learning is the tool for marketers to drive revenue and increase market share. The adoption of machine learning does not happen overnight. We will discuss the Five Es of machine learning maturity – Educating, Exploring, Engaging, Executing and Expanding. Hear real-world examples of using machine learning to accelerate revenue, identify new customers and introduce new products based on machine learning capabilities.
Next DSS MIA Event - https://datascience.salon/miami/
Today there is a lot of buzz around customer experience. Many companies have realized that investments in customer experience improvement is important not just because it helps to boost the bottom lines of their businesses but because it takes at least 4 to 6 times more cost to acquire a new customer than to retain an existing customer.
Accelerate Business Growth and Outcomes with AICognizant
Here’s how 10 organizations are using AI to accelerate decision making, improve processes, enhance user engagement, reduce costs and drive business performance.
Project Context:
Development of a business intelligence solution for a Tunisian pharmacy.
Achievements:
• Extraction, transformation, loading and analysis of data
• Data scrapping and storage
• Reporting and creation of Dashboards
Technical environment:
• MS SQL Server Integration Services (SSIS)
• MS SQL Server Analysis Services (SSAS)
• MS SQL Server Reporting Services (SSRS)
• MS Power BI
• Python
Data Analytics for E-Commerce: Driving Growth with Expert TrainingUncodemy
Data analytics has become an essential component of the e-commerce industry, driving growth and success for businesses all over the world. Organizations require data analytics professionals to harness the power of data and make informed decisions. This is when data analytics training comes in handy.
Monitoring Analytics To Create Customer Value And ExperienceeTailing India
According to research conducted by Gartner,Customer Experience (CX) is the top priority for companies who have invested in analytics software. The goal for any company is to have an ‘always on’ view of how their operational performance that impacts on the way that customersexperience their brand across all touch-points. This is now possible by using untapped machine data in combination with more traditional measures of customer satisfaction such as Net Promoter Score (NPS).
Machine learning and remarketing are two very popular ways of enhancing marketing campaigns. Used in tandem, they can deliver much better business outcomes. This session reveals how to get started with machine learning-driven remarketing using R.
1. Foundations of Business Analytics
The Body Workshop Case study
The client is a Gym-Fitness centre who is trying to establish an online presence. They would
like to use the emerging digital technologies to gain an edge over the competition. They seek
our advice on how to define Social Media Metrics to measure the success of their effort.
My team interviewed the Management of Finance, IT and Business Development - Strategy
departments (Roles played by Faculty members). The detailed business understanding was
obtained. We finalized their business Objectives based on which we arrived at the Data
Analytics Objectives. After conducting a thorough research study, I got to present my team’s
recommendations to the Panel.
Car Engine Performance Analysis Case study (Multivariate
Data Analysis)
The Data set consisted of different parametric attributes of cars sold in the US in the 50’s,
60’s, 70’s. By using various analytic techniques like Principle Component Analysis, Multi-
Dimensional Scaling, Cluster Analysis the cars were classified into different categories.
Based on these clusters, it is possible to identify different selling points of the cars to be built
in the future and those in the pipeline. By studying the Performance features and Fuel
Economy of the cars, we would be able to perform targeted marketing and effectively sell the
cars to different kinds of customers.
Employee Salary Prediction Case Study
The Data set contained various factors that determine the salary of employees in their
organization and their respective salaries. The Linear model we built would help determine
the salary of potential candidates based on their personal profile such as Years of relevant
experience, Number of subordinates they manage, Previous salary etc. We built a Multiple
Linear Regression Model to predict the salary of applicants.
Advanced Analytics
Minority Report – Insurance Fraud Detection Case Study
Given a Data set of Historical cases of Frauds identified by an Insurance service provider.
By detecting the patterns in the data, we built models that identified suspicious cases in a
new dataset and presented our findings to the panel.
Neural Networks model for Stock Price Prediction
The Data Set is the historical stock prices of the Straits Times Index (STI). The data was
obtained from Yahoo Finance. This data was used to predict the closing price, Opening
Price of the next trading session. The practical application of this is obvious. The prediction
can be used for intraday trading of the futures contracts of Straits Times Index (as an
indicator in Technical Analysis).
2. Association and Market Basket Analysis – Auto Desk
Website Case Study
Given Data is a list of Webpages a Potential Customer visited in a session of web browsing
in the Autodesk Website and the corresponding purchases they made. Based on this
information, we were required to predict the future purchase pattern of the Customers. The
Auto desk is seeking to build an effective product recommendation engine for its repeat
customers. Using APRIORI algorithm, Market Basket Analysis technique, Association
Analysis, the product recommendation engine was built.
Campaign Management
Google Analytics Project – Google Online Marketing
Challenge
We organized an online Marketing challenge for our client - Peppy Silver, a web based
jewellery retailer and wholesaler. We helped the client to achieve 60% increase in Site
Traffic and 150% increase in Sales Revenue in a period of 2 month period.
Targeted Marketing Campaign and Measuring Marketing
Campaign Effectiveness
A Bank needs to reach to a wider range of Customers for its fledging Credit Cards Business.
We were required to study the market conditions and recommend a suitable mode of
advertisement for the different customer segments. We had to manage the entire marketing
budget and allocate the optimal amount of money to the different concurrent campaigns.
Various Campaign success criteria were defined and the progress was observed throughout
the campaign lifecycle and suitable changes were recommended to increase the campaign
efficacy.
Paid Social Media Campaigns
A review was done on Social Media Goals and Metrics. The relative merits and demerits of
Traditional Marketing Vs Social Media Marketing was conducted as a comparative Analysis.
Computational Intelligence
Performance Analysis of Hybrid Neural Networks
Two separate data sets were provided (Diabetes data, Wine Quality)
Classification Model was built using Neural Networks Algorithm. The Diagnostic Data shows
patients showing signs of diabetes as described by WHO. The model once trained can
predict and classify the Test set of the data into Diabetes Patients and non-Diabetic. Various
tools (R/Rattle, Neuroph, Neuro solutions, Neural Tools – Palisade, Rapid Miner student
version) were used to build a Hybrid Model.
Regression Model was built to estimate the wine quality. The given dataset contained the
physio chemical composition of the white variants of the Portuguese “Vinho Verde” Wine.
3. The linear model we built would predict the quality of the wine based on the test results of
the Physio Chemical Test results.
Support Vector Machine models were studied in detail, Literature survey
was summarized and we presented our understanding with suggestions of practical
applications of SVM.
Customer Relationship Management
Connected Hardware Case Study
Product Development Strategy – Conducted a detailed Market Research (Studying Target
Customers, existing competition, the market share of the competitor, penetration strategy)
for a New Product launch, Analysis of the Product Lifecycle, Pricing Strategy, Formulating a
Business Model, Partnering with Establishment, Planning a Merger with Technology
Hardware manufacturer, creating a value chain.
Recency – Frequency – Monetary Analysis: Studying Customer
purchase patterns
Customer Lifetime Value: Determining the maximum viable expenditure (Customer
Acquisition Cost) that can be incurred to optimize the profitability of the business.
Health Care Analytics
Framingham Heart Study
This Case study involved identification of Risk Factors, assessment of their predictive ability,
implications of disease prevention. I presented our research study to the review panel on
behalf of my team.
Brand Analytics
Understanding the importance of Branding, Calculating Brand Value, and Building Brand
Equity for an enterprise. Understanding significance of Branding in the Healthcare and
Pharmaceutical Industry.
A study on Clinical Trials, Survival Models
Understanding and interpreting the results of Cox Regression models.
Data Analytics
World Economic Forum – Visualization Project
The topic I chose to work on: An E-Commerce start-up hotspot in South East Asia. The
Economic growth, Infrastructural, Administrative, Technological Innovation, Ease of doing
Business and few other important factors were studied.
The data from WEF was used to develop Visualization that targets Government officials in
ASEAN countries as to what are the most important areas to be focused on in order to
attract foreign investments in their countries.
4. The Visualization would also help existing E-Commerce companies to expand their business
to neighbouring countries by highlighting opportunities, it would also act as a leading
indicator for Investors who are looking to grow their capital by investing in emerging
enterprises in the ASEAN region.
Ensemble Model
Our client has a Bicycle renting company. Their business is to procure bicycle from a
provider at a low cost and rent it out to cyclists who can use it for one day and return it. The
problem that needs to be addressed is if the client procures more cycles than the following
day’s demand, the procured cycles go underutilized and have to be paid for anyway. If the
demand is not met with enough procurement, they risk losing business to competition.
Hence, we have been called to predict tomorrow’s demand based on historical demand.
Since the available data was only 18 months daily transactional data, we were unable to use
Time series model. Therefore we decided to use the Ensemble of different model. Bagging,
Boosting techniques were used side by side with Moving Averages. We produced an
increase of 18% in profits compared to Naïve Bayes techniques that was previously used to
predict tomorrow’s demand.
Decision Making and Optimization
Multi-criteria Optimization – Bio Mimetic Batoid Robots
design Automation project.
An attempt has been made to deploy a large number of autonomous batoid robots in the
coastal region in hope that they can freely and independently move around and study the
ocean bottoms without affecting the eco system. This effort has received several accolades
from environmental groups for its non-invasive and sustainable scientific methodology.
Robots that resemble the Sting ray have been constructed using 3D printer and powered by
battery have been designed and deployed as an experiment. These batoid robots would be
able to move using the Actuators on each side of their fins. Every aspect of the robot is
currently being tested by the method of trial and error.
If and when the robot fails during the testing phase the scientists gather around the drawing
board to start from scratch. Once they figure out what went wrong, they model a new design
and 3D print the robot with modifications. Thus several iterations later they have a decent set
boundary conditions that define the solution space.
Here we find an opportunity to intervene in the indiscriminate process of random trial based
experimentation of new designs for the future iteration of designing the robot. Instead, if we
are able to optimize the search space of the problem using Linear Programming, the choice
of parameter values in the future iterations would be falling within the well-defined feasible
region.
In an ideal scenario, the Batoid would look and behave exactly like a Sting ray. But, the
reality is more complex. Changing one parameter has found to affect other aspects of the
robot.
We successfully framed a simplistic problem statement to help address the above said issue
and using Multiple Criteria Optimization techniques learnt.
5. A cost-benefit analysis of a potential third casino operator
license to local, foreign or joint venture in Singapore
Since the two integrated resorts (IRs) opened in 2010 in Singapore, they have proven to
contribute significantly to the tourism growth. The two casinos are also among the most
profitable casinos in the world. The two IRs support, directly and indirectly, more than 40,000
jobs economy-wide. All these have been achieved while keeping the law and order risks to
our society tightly contained and establishing strict social safeguards to protect the local
population.
Our objective was to determine the feasibility of a Third casino in Singapore while evaluating
the relative merits and demerits of issuing license to a Local operator.
Text Mining
Analysing Accident reports to determine most major cause of accidents in workplace
Using Text Mining techniques, we determined the reason behind the workplace related
mishaps. The result of this activity could help the employers take necessary precautions to
keep the work environments safe for their employees.
The employer can also include important provisions in the insurance terms based on the
insights we generate.
The local hospitals can be better prepared to treat patients who succumb to injury during
work and successfully treat them before it results in major complications.
Logistics and Supply Chain Analytics
1) Matching Factory Produce and Demands in different cities
2) Rolling Forecast, MAPE, Safety Stock, Reorder Point, Reorder Quantity estimation
3) Fleet Routing and Scheduling of a Trucking Company in Singapore using Routific
Tool.
4) Warehouse Planning and Global Supply Chains
Web Analytics
Recommendation Engine for Movies
The dataset was derived from Movie Lens website. It gives in complete detail all the movies
that have been watched and rated by thousands of viewers, based on their past preference,
using Association Analysis, we built a recommendation engine that predicts the movies that
they would love the most. We used the APRIORI Algorithm and the CARMA node to build
this Engine.
Identifying Baby Whales in Research Communities
The given data contained the Publications, Co-Authorship of various members of the
scientific community. While it is easy to identify domain experts based on the number of
Citations they have, Number of publications they have in Journals with high impact factor,
this information does not help us. This is because the domain experts are not easily
approachable. They tend to demand a hefty consultation fee and cannot be easily hired due
to their prior commitments with some on-going projects.
6. Hence, it becomes very essential to identify “baby whales” who are equally brilliant but yet to
reach their peak of achieve. These baby whales are promising candidates in their chosen
field and are much more easily approachable. They can be hired for a nominal pay and can
hit the ground running once hired.
We used the Link Analysis to identify Co-Citations, Page Rank Algorithm, HITS to detect
Rising stars and community detection in the domain. This helps us to identify a well-
established team with members of similar interest and expertise.
Our results would help Entrepreneurs hire early stage employees for R&D roles in their start-
up. I presented our finding to the assessment panel on behalf of my team.