Customer Analytics Using Machine Learning Training Course

Duration 10 days 17 modules · 65 contact hours
Level Intermediate Certificate on completion
Who attends Customer Analytics Manager, Data Analyst and others
Fee from USD 2,200 KSh 190,000 for Kenyan organisations
Where 52 cities + online See dates by city
Next start 16 Nov 2026 Course code SD-DS-052

Before you join: Participants should be comfortable working with spreadsheets or tables of customer data and interpreting basic business measures such as conversion rate, retention rate and revenue.

Dates & prices

Two start dates a month in every city, and live online on weekdays or weekends. Fees are per participant; classroom fees include tuition, materials, lunch and refreshments, and the certificate.

Suggest a date or city

Showing the next six months. Later dates, other cities and private deliveries for your team are available on request.

Course overview

Customer data is often spread across CRM systems, transaction platforms, digital channels and service records. Teams may know who bought, clicked or complained, but struggle to identify which customers are likely to leave, respond to an offer, increase value or require intervention. Machine learning can turn these records into scored customer decisions, provided the data, model choice, evaluation approach and deployment rules are fit for purpose.

This course teaches participants to design and build customer analytics models for acquisition, retention, cross-sell, customer value and campaign prioritisation. Participants use Python, SQL and scikit-learn to prepare customer-level datasets, engineer behavioural features, build classification and regression models, segment customers, estimate churn risk and assess campaign uplift. The course also covers customer lifetime value, recommendation methods, model explainability, bias checks, monitoring and responsible use of personal data.

Teaching combines concise technical briefings with guided analysis of realistic banking, retail, telecommunications, insurance and digital-service cases. Participants work with customer datasets from the first week, progressively improving a reusable analytics project. They leave with a documented customer analytics model pack containing a business problem statement, feature dictionary, model evaluation report, customer score output, action rules and deployment plan.

The course suits analysts and managers who need to move from descriptive customer reporting to repeatable, evidence-based customer decisions. It is particularly relevant where commercial, service and risk teams need transparent models that can be explained to business owners, compliance teams and operational users.

What you will be able to do

  • Frame customer growth problems as measurable machine learning use cases
  • Build customer-level analytical datasets from transactional, behavioural and CRM records
  • Engineer recency, frequency, monetary and behavioural customer features
  • Train and compare classification, regression, clustering and recommendation models
  • Evaluate churn, propensity and customer value models using business-relevant metrics
  • Interpret model predictions using feature importance, SHAP values and decision rules
  • Design customer treatment strategies using scores, segments and uplift evidence
  • Produce a documented customer analytics model pack and deployment plan

Benefits of attending

For you

  • Build a portfolio-ready customer churn, propensity or value modelling project
  • Translate commercial customer questions into measurable analytical model specifications
  • Explain predictive model results clearly to marketing, service and senior stakeholders
  • Select model metrics that reflect campaign economics and customer treatment capacity
  • Create documented analytics outputs suitable for review, handover and reuse

For your organisation

  • Prioritise retention activity using predicted churn risk and expected value
  • Reduce untargeted campaign spend through propensity-based customer selection
  • Improve offer relevance through behaviour-based segments and recommendation logic
  • Establish consistent controls for customer model validation, fairness and monitoring
  • Create reusable data, feature and model documentation for future customer initiatives

Course modules

17 modules over 10 days

Participants can define a customer analytics use case with measurable commercial and operational success criteria.

  • Customer lifecycle decisions: acquisition, activation, retention, growth and win-back
  • Problem framing with decision, population, prediction horizon and action
  • Churn, propensity, value and next-best-action use case distinctions
  • Business value trees linking model outputs to revenue, cost and risk
  • Defining intervention capacity, contact policy and treatment constraints
  • Success criteria using incremental profit, retention and customer experience measures

Participants can specify the data structure required for customer-level machine learning analysis.

  • Customer analytical record defined by entity, observation window and outcome
  • Identity resolution across CRM, transactions, web events and service cases
  • Fact tables, dimension tables and customer master data relationships
  • Choosing customer, account, household or device analytical grain
  • Point-in-time data snapshots preventing future information entering training records
  • Data lineage documented from source system through model-ready dataset

Participants can assess customer data quality and identify key governance controls before modelling.

  • Completeness, validity, consistency, timeliness and uniqueness data quality dimensions
  • Missing-data mechanisms: MCAR, MAR and MNAR practical implications
  • Duplicate customer records and survivorship rules for master data
  • Personal data minimisation, purpose limitation and access-control requirements
  • Sensitive attributes, proxy variables and prohibited model input review
  • Data quality scorecards prioritising remediation by analytical business impact

Participants can prepare an analysis-ready customer dataset using SQL and Python.

  • SQL joins combining customer profiles, transactions and interaction histories
  • Window functions calculating rolling customer activity and purchase measures
  • Python pandas transformations for dates, categories and missing values
  • Aggregating event-level records into customer-level analytical features
  • Train, validation and test partitions created using time-aware splits
  • Reproducible notebooks separating extraction, transformation, modelling and reporting

Participants can identify behavioural patterns, data issues and candidate model signals through exploratory analysis.

  • Distribution profiling for customer tenure, spend, frequency and engagement
  • Cohort analysis comparing retention patterns by acquisition month
  • Funnel analysis locating drop-off across customer journey stages
  • Cross-tabulations revealing response differences among customer groups
  • Outlier inspection for extreme transactions, balances and service activity
  • Visual diagnostics connecting customer behaviour to target outcomes

Participants can create behavioural features that represent customer activity, change and relationship depth.

  • RFM features calculated from recency, frequency and monetary value
  • Rolling windows capturing recent activity, trend and behavioural volatility
  • Customer tenure, lifecycle stage and relationship-depth feature construction
  • Channel preference features derived from digital, branch and contact-centre interactions
  • Lagged measures preventing target leakage in predictive customer models
  • Feature dictionaries documenting definitions, sources and expected refresh schedules

Participants can build and interpret customer segments for differentiated service and marketing actions.

  • Segmentation objectives: needs-based, value-based, behavioural and lifecycle segments
  • Standardisation and transformation before distance-based clustering methods
  • K-means clustering using inertia, silhouette score and cluster profiles
  • Hierarchical clustering for examining nested customer group structures
  • Principal component analysis for visualising high-dimensional customer features
  • Segment naming based on observable needs, value and treatment opportunities

Participants can design a churn prediction model with a valid target, horizon and intervention window.

  • Churn definitions for subscription, banking, retail and digital-service customers
  • Prediction windows aligned with retention campaign lead times
  • Binary target construction from cancellation, inactivity or relationship closure
  • Class imbalance diagnosis using prevalence, confusion matrices and class weights
  • Baseline churn models using logistic regression and simple business rules
  • Retention eligibility rules separating predictable churn from actionable churn

Participants can train and compare classification models for response, conversion and customer risk outcomes.

  • Logistic regression coefficients interpreted as customer outcome associations
  • Decision trees producing readable split rules for customer selection
  • Random forests reducing variance through multiple decorrelated decision trees
  • Gradient boosting models for nonlinear customer response patterns
  • Cross-validation estimating model performance across alternative training samples
  • Hyperparameter tuning using grid search and validation performance

Participants can estimate customer value using regression methods and practical lifetime value assumptions.

  • Customer lifetime value components: margin, retention, frequency and discount rate
  • Regression targets for future spend, margin and transaction frequency
  • Linear regression assumptions, residual checks and coefficient interpretation
  • Regularisation with ridge and lasso regression for correlated features
  • Mean absolute error and RMSE for forecast accuracy assessment
  • Value tiers combining predicted value with confidence and actionability

Participants can evaluate customer models using statistical performance measures and commercial decision criteria.

  • Confusion matrix interpretation for retention and campaign targeting decisions
  • Precision, recall, specificity and F1 score trade-offs
  • ROC curves and AUC for ranking customer risk predictions
  • Precision-recall curves for rare response and churn events
  • Decile lift charts ranking customers for limited treatment capacity
  • Threshold selection using contact cost, expected margin and intervention capacity

Participants can distinguish likely responders from customers whose outcomes change because of treatment.

  • Propensity modelling versus uplift modelling for campaign decisions
  • Treatment and control groups designed for causal customer measurement
  • Incremental response calculated from treatment-control outcome differences
  • Four uplift segments: persuadables, sure things, sleeping dogs, lost causes
  • Uplift decision trees for treatment-effect customer segmentation
  • Campaign holdouts protecting valid measurement of incremental commercial impact

Participants can select recommendation approaches suited to available customer, product and interaction data.

  • Rule-based next-best-action using eligibility, suitability and contact constraints
  • Collaborative filtering from customer-product interaction matrices
  • Content-based recommendations using product attributes and customer preferences
  • Association rules identifying frequently purchased product combinations
  • Cold-start strategies for new customers, products and sparse histories
  • Recommendation evaluation using precision at k and business guardrails

Participants can explain model outputs and identify fairness risks in customer decision processes.

  • Global feature importance identifying primary model prediction drivers
  • SHAP values explaining individual customer scores and model behaviour
  • Partial dependence plots showing average feature-outcome relationships
  • Fairness checks comparing selection rates across protected customer groups
  • Proxy discrimination risks arising from location, language and behavioural variables
  • Reason codes translating model outputs into reviewable customer decisions

Participants can plan controlled deployment, monitoring and retraining for a customer analytics model.

  • Batch scoring schedules aligned with campaign, service and sales processes
  • Model output tables containing scores, segments, reason codes and timestamps
  • Champion-challenger testing for controlled model replacement decisions
  • Data drift and concept drift monitoring indicators
  • Performance dashboards tracking lift, conversion, retention and realised value
  • Model documentation covering assumptions, limitations, approvals and ownership

Participants can define roles, controls and working routines for sustained customer analytics delivery.

  • RACI matrix for data owners, analysts, marketers and model approvers
  • Backlog prioritisation using business value, feasibility and data readiness
  • Experiment calendars coordinating campaigns, holdouts and model refreshes
  • Model risk controls for validation, change approval and audit evidence
  • Stakeholder reporting tailored for executives, campaign teams and technical reviewers
  • Analytics adoption measures tracking use of scores in frontline decisions

Participants can build, review and present a complete customer analytics model pack for a defined growth decision.

  • Select a customer growth problem with measurable decision and outcome
  • Prepare customer data and document analytical grain, windows and quality checks
  • Engineer features and train a justified predictive or segmentation model
  • Evaluate commercial performance using lift, thresholds and treatment capacity
  • Present score outputs, action rules, fairness checks and operating controls
  • Peer review model pack against business, technical and governance criteria

Exercises and examples are adapted to each cohort's sectors and roles.

What you take back to work

  1. Customer analytics use-case canvas
  2. Customer feature dictionary template
  3. Churn or propensity model notebook
  4. Model evaluation and threshold report
  5. Customer treatment decision matrix
  6. Customer analytics deployment checklist

Who should attend

  • Customer Analytics Manager — needs repeatable models for retention, growth and service decisions
  • Data Analyst — wants to build predictive customer models from operational data
  • CRM Manager — needs evidence for targeting, contact strategy and campaign prioritisation
  • Marketing Analyst — wants to improve response, conversion and offer selection
  • Data Scientist — needs customer analytics methods and commercial evaluation measures
  • Product Manager — needs customer evidence to prioritise journeys, offers and interventions

Skills and tools covered

Customer data preparationChurn predictionCustomer segmentationPropensity modellingModel evaluationAnalytics deployment planning

Tools and standards

Python, SQL, scikit-learn, SHAP

How the course runs

Training methodology

The course uses short instructor briefings followed by guided work in Python notebooks and SQL-based customer data tasks. Cases cover churn reduction, campaign response, next-best-action and customer value decisions from banking, retail, telecommunications and subscription services. Participants inspect data quality, build models, compare results and discuss treatment choices in groups. Simulations test how threshold selection, contact costs and capacity constraints affect commercial outcomes. Each participant develops an application plan and model pack relevant to their own customer data context.

Requirements and prerequisites

Participants should be comfortable working with spreadsheets or tables of customer data and interpreting basic business measures such as conversion rate, retention rate and revenue. Prior experience with SQL, Python or another analytical programming language is helpful, because examples use Python notebooks, but it is not assumed at an advanced level. Participants should understand basic descriptive statistics, including averages, percentages and distributions. Prior machine learning experience is not required. The course does not require advanced calculus, software engineering experience or prior use of cloud machine learning platforms.

A typical training day

08:30 – 10:30First session
10:30 – 10:45Refreshment break
10:45 – 12:30Second session
12:30 – 13:30Lunch and networking
13:30 – 15:00Third session
15:00 – 15:15Refreshment break
15:15 – 16:30Workshop and daily review

Live online deliveries follow the same structure in East Africa Time, with shorter screen blocks and longer breaks.

What the fee includes

  • Instruction by a practitioner facilitator
  • Full course workbook and materials
  • Exercise files, templates and case studies
  • Refreshments and lunch on classroom deliveries
  • Certificate of completion
  • Post-course application plan

Frequently asked questions

Participants should be comfortable with tabular data, business metrics and basic descriptive statistics. Some Python or SQL familiarity is helpful, but advanced programming and prior machine learning experience are not required.

Bring a laptop capable of running a modern web browser and Python notebook environment if attending in person. Participants may also bring a non-sensitive example of a customer analytics problem or data dictionary from their workplace.

The course is designed for analysts, CRM and marketing professionals, customer insight teams, product managers and data scientists. It is most useful for people responsible for retention, campaign targeting, customer value or next-best-action decisions.

General machine learning courses often focus on algorithms across many problem types. This course concentrates on customer data, lifecycle decisions, campaign economics, treatment capacity, customer fairness and model deployment into CRM or service processes.

Participants can use the templates and methods to scope a churn, response, value or segmentation initiative using their own customer data. The model pack structure also supports discussion with marketing, data governance, compliance and technology stakeholders.

Each training day runs from 08:30 to 16:30 and provides about 6.5 contact hours. The first week establishes customer data, features and initial models; the second week covers commercial evaluation, explainability, deployment and the final capstone presentation.

Questions about this course?

Ask about dates, fees, group bookings or running it for your team. Our training team replies by email.

Ask on WhatsApp info@skillsetdevelopment.com

More courses in Data Science

Data Science Big Data Analysis with R and Python Training Course 10 days · from USD 2,200 · next 16 Nov Data Science Big Data Analytics with Python Training Course 10 days · from USD 2,200 · next 16 Nov Data Science Data Analysis and Machine Learning Using R Training Course 10 days · from USD 2,200 · next 16 Nov Data Science Data Analysis and Visualisation with R and Python Training Course 10 days · from USD 2,200 · next 16 Nov Data Science Data Science and Machine Learning Training Course 10 days · from USD 2,200 · next 16 Nov Data Science Data Science and Predictive Modelling Training Course 10 days · from USD 2,200 · next 16 Nov Data Science Spatial Analysis Using R Training Course 10 days · from USD 2,200 · next 16 Nov Data Science Time Series Analysis Using R Training Course 10 days · from USD 2,200 · next 16 Nov