Marketing Analytics: Chapter 1

Chapter 1: Introduction to Marketing Analytics

• Marketing involves managing customer needs, making strategic decisions about products, pricing, distribution, and communications.
• Marketing uses insights from marketing analytics to satisfy customers and maintain a competitive advantage.
• Prices are updated based on demand via algorithms and technology (Hotels.com, Expedia, Orbitz).
• Spotify uses user-generated playlists, listener preferences, and data analytics to suggest songs.
• Stitch Fix combines stylist input with data analytics algorithms for personalized clothing suggestions.
• Marketing Analytics uses data, statistics, mathematics, and technology to solve marketing problems by modeling and software
• Has become more accessible due to large data availability, improved techniques, and increased computer processing power.
• Driven by user-generated data from social media, mobile apps, and shopping channels like pricing, product development, channel management, and communications.
• Roy Rogers uses analytics for site selection and sales forecasting
• Higher search volume compared to financial, HR, and supply chain analytics needing analytics to bridge the gap due to being "data rich but information poor".
• It will become ubiquitous innovative processes needing problem-solving, communication, collaboration, project management
• Descriptive Analytics: explains or quantifies the past (e.g., data queries, visual reports, descriptive statistics).
• Predictive Analytics: forecasts future trends.
• Prescriptive Analytics: recommends actions.
• AI and Cognitive Analytics: advanced techniques for deeper insights.
• Descriptive Analytics: A set of techniques used to explain or quantify the past gathering customer feedback and improve experiences.
• Predictive Analytics: builds models based on past data to predict future values, needs, and opportunities.
• Target predicts a teen girl's pregnancy using customer purchase data and predictive modeling.
• Zillow uses over 100 predictors to estimate home values ("Zestimates") for millions of homes.
• Prescriptive Analytics identify the best optimal course of action or decision as demonstrated by the example of UPS optimizing delivery routes to reduce left turns.
• Airlines optimize flight and crew scheduling to reduce costs and increase satisfaction.
• Amazon uses price optimization, changing prices frequently to influence customer behavior.
• Kellogg uses optimization models for production and supply chain management, saving over $475 million annually.
• AI and Cognitive Analytics mimics human-like intelligence for tasks such as pattern recognition, image recognition, text understanding, and voice processing.
• Olay uses image recognition to recommend skincare products, doubling sales conversion rates.
• Mtailor uses customer images to create customized clothing fits, claiming higher accuracy than professional tailors.
• Hitachi uses AI ("H") to discover patterns and improve operations in various business functions.
• Defining he Right Business Problems is a critical initial step in the marketing analytics process needing to understand customers’ journey, identify opportunities, define business question .
• A retailer facing bankruptcy needs to entice first-time mobile app users to make purchases can be solved with retailer facing bankruptcy needs to entice first-time mobile app users to make purchases.
• Involve stakeholders in determining project requirements.
• Feedback collection methods interviews, observation, surveys, and brainstorming sessions are important.
• The traditional six discovery questions of What, Who, Where, When, Why, and How are useful for defining the business problem.
• SMART Analytics Principles: Specific, Measurable, Attainable, Relevant, and Timely.
• Focus on understanding data sources for data requirements.
• Primary Data is collected for a specific purpose using methods such as surveys, focus groups, interviews, observations, experiments.
• Walmart uses facial recognition to detect unhappy customers and provide assistance.
• Secondary Data is existing data collected for another purpose used for formulating ideas and exploring current business questions.
• Public Datasets: Google Dataset Search includes data from NASA, NOAA, Harvard's Dataverse, GitHub, Kaggle.
• Online Sites data includes browsing behavior, purchase history, social media chatter.
• Mobile Data tracks customer behaviors and locations
• Channel data is shared among suppliers, wholesalers, distributors, retailers.
• Commercial Brokers like Acxiom aggregate and sell data (e.g., socioeconomic status, health interests, political views).
• Corporate Info: Data from business transactions and customer interactions across functional areas.
• Government Sources include federal data collection from local, state, and federal agencies.
• Data.gov has over 200,000 datasets searchable by topic.
• Consumer Complaint Data has customer sentiments about financial products and services.
• Demographic data is organized By ZIP code, gender, ethnicity, and citizenship.
• Department of Agriculture data covers prices for over 153 commonly consumed fruits and vegetables.
• Tax Return Data is broken down by state and ZIP codes levels.
• Structured data is organized in rows and columns stored in databases or spreadsheets used in various analytics techniques.
• Unstructured data includes text, images, videos, and sensor data needing advanced analytics techniques increasingly collected due to tech
• Cisco Systems and Dell Technologies combine unstructured data via graph databases facilitating integration
• Data Measurement categorizes data based on type and means of collection
• Numerical Data: Quantitative measurements.
• Categorical Data: Qualitative classifications.
• Discrete Data: Whole numbers.
• Continuous Data: Values with decimals.
• Binary: Two values (e.g., yes/no, 1/0).
• Nominal: No meaningful order (e.g., marital status, country).
• Ordinal: Natural order with uneven intervals (e.g., product preference rankings).
• Interval Scales: Equal intervals, no absolute zero (e.g., temperature in Fahrenheit).
• Ratio Scales: Equal intervals, absolute zero (e.g., sales revenue, weight).
• Independent Variables influence or drive the dependent variable.
• Dependent Variables is the outcome influenced by predictor variables.
• Supervised Learning: Target variable is known and available in a historical dataset (labeled data).
• Training Dataset: Used to build the algorithm and learn the relationship between predictors and the target variable.
• Validation Dataset: Assesses how well the algorithm estimates the target variable and selects the best model.
• Testing Dataset evaluates the final selected algorithm's performance on a third dataset.
• Prediction occurs when the target variableis continuous
• Classification happens when the target variable is categorical.
• Unsupervised Learning: No target variable, analyzes underlying structure and distribution in data.
• Association Analysis recommends products based on past purchases (e.g., Amazon's "others who bought this item also bought").
• Cluster Analysis groups customers into segments based on loyalty and purchasing behavior.
• Use unsupervised learning to determine customer loyalty segments, then apply supervised learning to predict purchase amounts for each segment.
• The 7-Step Marketing Analytics Process includes identifying the business problem, data collection, data preparation, model development, model evaluation, communication, and deployment.
• Data Preparation and Feature Selection involves combining, examining, cleaning, merging, and refining.
• Modeling Development's method selection depends on target variable type and business question.
• Evaluation involves key stakeholder collaboration and model communication.
• Approval: Management must approve deployment.
• Ensure Privacy, Accuracy, and Ethics.
• Understand data sources and limitations is important which can be helped through Abraham Wald's analysis of bomber damage vs modern testing

Chapter 4: Data Visualization

• Data visualization is the ability to communicate practical insights to a wide audience.
• It is an effective way to quickly and clearly convey insights when data provides value only when it is quickly and easily accessible.
• Visualization tools like Tableau help create compelling "data stories" combining data analysis with computer graphics that encodes quantitative values into graphical formats to improve communication.
• Companies use data visualization for better decision-making like Coca-Cola and Walmart partnership sharing product data in Tableau format.
• This resulted in a $20 million reduction in lost sales over 13 weeks by improving in-stock availability.
• Leads to efficient and accurate decision-making and a data-driven cultural change using P&G's Decision Cockpits
• Data Visualization allows for exploration and comparing results efficiently via geographic display
• Important to engage the audience through simple and clear communication while highlighting trends and patterns, reduce audience fatigue.
• Visualize with Tableau, Google Data Studio, Qlik, Microsoft Power BI for descriptive analysis and advanced functionality.
• Design principles improve perception, attention, interpretation, comprehension, and memory of visual data.
• Balance objects using symmetrical and asymmetrical methods for chart visualization.
• Using emphasis to highlight important insights through color, sizes etc
• Proportion the data through the size of objects
• Aim for seamless design element transitions with complementary colors
• Variety to use different visualizations to engage and aid
• Aim for harmony with all elements, removing anything which does not contribute to the overall story.
• Aethetics are a vital piece of engaging or dismissing audiences, vital to balance with clarity.
• Critical to set mood, emphasize points, and guide depending on the type of graph and using intuitive colors
• Overcomplicating visuals can lead to confusion and loss of context.
• Simple designs are often more effective than complex ones.
• Recommendations: Use position, length, and slope for easier understanding.
• Avoid: pie charts, 3D charts, and charts with two y-axes as they are hard to read, opt for simpler solutions.
• Use numeric data while understanding common charts and graphs.
• Business data is data for costumers, etc with relationships displayed various ways.
• Time is used for value evolving and line arts are effective for changes through time.
• Product prices and locations are used for rankings. Part-to-Whole Relationships are the same as ratios ie FedEx revenue from Amazon.
• Correlations with 2 data sets like Scatter plot allow positive and negative correlations.
• Frequency Distributions indicate how many observations fall within a certain interval.
• Geographic charts display food safety issues, customer purchase locations, and location of feedback
• Deviation Analysis is standard referencing the difference between Uber and Lyft.
• Data simplifies for diverseaudiences and enables you to communicat

Chapter 5: Regression Analysis

• Regression Modeling captures the relationship strength between a single numerical dependent variable (Y) and one or more predictor variables (X).
• To predict customer purchase spending look at email promotion and income level as a component.
• Y variable being purchase amount and X promotion plus income.
• Simple Linear Regression examines the relationship between a single dependent variable and a single independent variable.
• Look at the relationship between home size and sales price.
• Equation: y=b0+b1x+e/ Estimated y intercept/ Slope of regression and error term Methods: Ordinary Least Squares (OLS) minimizes the sum of squared errors process The model will calculated weights and find the line that makes it zero. R² is the amount of variable explained by the independent variables that if close to 1 means better prediction.
• Appropriate for million dollar investing, turning customers, and predicting sales and seeing effect of sales in ecommerce.
• Three types are used depending on whether its between or within models
• Model analysis looks at relationship between customers and advertising to improve them.
• Types of Regression Models: Descriptive, Explanatory, and Predictive.
• Training, Validataion, and Testing is run through the model.
• Helps and makes a strong prediction.
• Descriptive/Explanatory focuses on causation and retrospective methods.
• The model uses to build it with signfiance and goodness and fit.
• Predictive uses a prospective new observation that divides and compares it with a data set.
• Validate via high accuracy in observations or data.
• Mean Absolute Error (MAE): Measures the absolute difference between predicted and actual values.
• Mean Absolute Percentage Error (MAPE): Percentage absolute difference between prediction and actual target.
• Root Mean Squared Error (RMSE): Indicates how different the residuals are from zero.
• Need quality data to get a greater predication.
• See models through MAE as RMSE depending on test data.
• Valodation is built using a training test and then validated.
• Hold out valodaion happens with training data and testing separate data.
• Optional test sample happens with sample.
• Modeling of data comes down to handlinv categorical with dummy coding with O AND ! values.
• k-1 is used to avois linear combination error with flash sales and holiday promoed.
• B0 mean of sales B1 Difference between mean of coupon sale B2 Difference with holiday promotion
• Simplify your selections. Simplify with data Less variables to not overfit. Occurs and is more complex
• Mitigate as some by predictive mode.
• Run math examince
• Detect by adding variables compare with and without.
• Identify features to explain what variable is.
• Analyze through knowledge and quantitatively.
• Use all data to show improve performance.
• Eliminating less variables to improve the accuracy that lead to less errors. Statistical Analysis: Using software tools.
• Backwards: Start w all predictors then remove using significane step
• Foreward: create models for each model
• Stepswise: Combione forewords with backwards that is common for testing models.
• SPSS and etc

Introduction to Marketing Analytics covers customer needs, strategic decisions, and competitive advantage. Algorithms and technology update prices based on demand. Data, statistics, mathematics, and technology solve marketing problems.