Data Science
Master applied machine learning and data analytics across finance and energy. Learn Python, modeling, visualization, and MLOps through hands-on case studies and real datasets.
From analysis to prediction
Analytics describes what happened; data science predicts what will. This program makes that leap, building from Python, Pandas, and exploratory analysis up through the machine-learning core and into applied modeling for finance and energy. The emphasis throughout is applied: every technique is taught against real datasets and real problems, so you learn modeling as something you do, not just something you understand.
The program is careful to keep the fundamentals in view even as it reaches advanced techniques. Clean data, honest evaluation, and clear communication matter as much as the algorithms, because a model that is not properly validated or cannot be explained is worse than useless in a real setting. This grounding is what distinguishes a practitioner from someone who can only run a tutorial.
The applied emphasis is deliberate, because the gap between running a tutorial and solving a real problem is where most learners stall. Working every technique against real finance and energy data builds the judgment to handle messy, ambiguous situations, which is what actually gets people hired as data scientists.
The machine-learning core
At the center of the program is the machine-learning workflow: supervised learning for regression and classification, Scikit-learn workflows, feature engineering, and the model evaluation and tuning that turn a first attempt into a reliable one. These are the skills that recur in every data-science role, taught with the rigor that makes results trustworthy.
From there the program moves into applied modeling that reflects the firm's domains: predictive analytics for finance and energy, time-series forecasting with ARIMA, and deep learning for sequences with LSTM. Working these on real problems, P&L forecasting, churn, load forecasting, renewable performance, is what makes the techniques concrete and the portfolio credible.
The machine-learning core is taught with an insistence on honesty, split before you train, choose a metric that fits the problem, because these disciplines are what separate a trustworthy model from an impressive-looking but useless one. That rigor is more valuable, and rarer, than knowledge of any specific algorithm.
Toward production
A model in a notebook is not yet valuable; it has to run somewhere and keep working. The final module introduces the path to production: TensorFlow basics, packaging and serving a model, an end-to-end MLOps introduction, and the monitoring and retraining that keep a model honest as the world changes. This production awareness is increasingly what employers look for beyond modeling skill.
The program closes where analytics always should, with communication: explaining what a model does, what it does not, and what decision it supports, to stakeholders who are not data scientists. A model whose results cannot be communicated will not be used, so this is treated as an essential skill, not an afterthought.
The production and communication emphasis reflects where the field has matured. Employers increasingly want data scientists who can ship and explain a model, not just fit one in a notebook, and the program builds exactly that end-to-end capability.
See the method, not just the topic
A representative worked example from the program, so you can see the level of concreteness the curriculum works at.
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import roc_auc_score
# Split first, so evaluation is on data the model never saw.
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.25, random_state=42, stratify=y
)
model = LogisticRegression(max_iter=1000)
model.fit(X_train, y_train)
# For risk scoring, AUC matters more than raw accuracy because
# the classes are imbalanced (few defaults among many loans).
proba = model.predict_proba(X_test)[:, 1]
print("AUC:", round(roc_auc_score(y_test, proba), 3))
# The lesson is not the model but the discipline: split before
# training, and choose a metric that fits the problem.The full syllabus
Four modules of five chapters each, sequenced so the material builds cumulatively. Each chapter carries a note on what it teaches.
Module 1Python for data science
- 01Python, Pandas, and NumPy foundationsPython, Pandas, and NumPy are the data scientist's daily tools. Fluency in these tools is the baseline for the field.
- 02Data loading, cleaning, and transformationLoading and cleaning data is where most real time goes. Clean inputs are the precondition for any real result.
- 03Exploratory data analysisExploratory analysis reveals what a dataset is really like. You cannot model well what you have not explored.
- 04Visualization for analysisVisualization is part of analysis, not just presentation. A chart during analysis often beats a model later.
- 05Data storytelling for decision-makingCommunicating findings is treated as a core skill. A model no one understands will not be used.
Module 2Machine learning core
- 06Supervised learning: regression and classificationSupervised learning covers regression and classification. These two problem types cover most of applied ML.
- 07Scikit-learn workflowsScikit-learn provides a consistent modeling workflow. A consistent workflow makes results reproducible.
- 08Feature engineeringFeature engineering often matters more than the algorithm. Better features often beat a fancier algorithm.
- 09Model evaluation and validationHonest evaluation is what makes a model trustworthy. Evaluation on held-out data is non-negotiable.
- 10Model optimization and tuningTuning turns a first attempt into a reliable model. Tuning is where reliability is won or lost.
Module 3Applied modeling
- 11Predictive analytics for finance and energyPredictive analytics is applied to finance and energy. The domains make the techniques concrete.
- 12Time-series forecasting with ARIMAARIMA is a workhorse for time-series forecasting. ARIMA remains a strong, interpretable baseline.
- 13Deep learning for sequences with LSTMLSTM brings deep learning to sequential data. LSTM handles patterns ARIMA cannot.
- 14Financial analytics: P&L forecasting and churnFinance analytics covers P&L forecasting and churn. Finance use cases anchor the modeling in value.
- 15Energy analytics: load forecasting and renewablesEnergy analytics covers load forecasting and renewables. Energy use cases show the breadth of application.
Module 4From model to production
- 16TensorFlow basicsTensorFlow basics open up deep learning. Deep learning starts with solid TensorFlow basics.
- 17Packaging and serving a modelPackaging and serving turn a notebook into a service. Serving is the step that makes a model real.
- 18End-to-end MLOps introductionAn MLOps introduction shows the path to production. MLOps is increasingly what employers look for.
- 19Monitoring and retrainingMonitoring and retraining keep a model honest over time. A model drifts unless it is watched and refreshed.
- 20Communicating results to stakeholdersResults are only useful when stakeholders understand them. Communication is the final, decisive step.
How the program is taught
The program is applied throughout, teaching every technique against real finance and energy datasets and culminating in two capstone models. The most effective approach is to build each model yourself end to end, from messy data to honest evaluation, because that full workflow is what real data-science work requires and what interviews probe.
It deliberately keeps fundamentals in view even at the advanced end, insisting on clean data, proper validation, and clear communication alongside the algorithms. Following that discipline is what separates a practitioner who can be trusted with a real problem from someone who can only reproduce a tutorial.
Where data science leads
Data science remains one of the most in-demand and versatile data careers, and the modeling and MLOps skills built here map directly to data-scientist, ML-engineer, and quantitative-analyst roles. Because the program is applied to real domains, its graduates can point to work that resembles the job rather than coursework.
Within the journey, data science is the modeling heart of the track. It builds on the SQL foundation and feeds directly into the two applied finance programs, where its methods are specialized to BFSI use cases, so it is both a destination and a springboard.
What makes this program different
Many data-science courses stop at fitting a model in a notebook; this one insists on the full path from messy data to a communicated, production-aware result. That end-to-end honesty is its distinguishing feature, and it reflects what employers actually need.
The applied, domain-grounded approach is the second distinction. Working real finance and energy problems, rather than toy datasets, produces judgment and a portfolio that generic courses do not, which is what turns knowledge into a hireable skill.
What you will be able to do
- Work fluently in Python, Pandas, and Scikit-learn
- Build, evaluate, and tune machine-learning models
- Forecast time series for finance and energy
- Take a model toward production with MLOps basics
- Communicate model results to decision-makers
Who should take it
- Aspiring and practicing data scientists
- Analysts moving into modeling and machine learning
- Engineers adding data-science skills
- Finance and energy professionals going quantitative
Tools and how they are used
The program centers on Python with Pandas, NumPy, and Scikit-learn, adds TensorFlow for deep learning, and uses Power BI and Tableau for communicating results. These are the mainstream data-science tools, taught as a workflow rather than a feature tour.
The emphasis is on using the tools correctly, splitting before training, choosing sensible metrics, validating honestly, rather than merely knowing their syntax. That disciplined use is what makes the resulting models trustworthy.
Common questions and how to prepare
A frequent question is how much math is required; the program teaches the modeling intuition and workflow you need without demanding a research background. Comfort with Python and the SQL from the foundations program is the most useful preparation, and the applied focus keeps the math grounded in problems.
The common pitfall is chasing complex models before mastering clean data and honest evaluation. Focusing on the fundamentals, and building the two capstones properly as portfolio pieces, is the way to get the most from the program.
How it fits the wider track
Data science is one of the two core-analytics programs and the predictive counterpart to business intelligence. It assumes the SQL foundation and pairs with BI to cover both the descriptive and predictive sides of the field.
It also feeds directly into the applied finance programs, where its methods are specialized to banking and markets, making it the pivot between general data science and domain-specific AI.
What you build and keep
Deliver two end-to-end models on real datasets: a credit-risk scoring model using logistic regression, and a portfolio-optimization study in Python using the Sharpe ratio and Monte Carlo simulation, each with clean data preparation, honest evaluation, and a clear write-up of what the model does and does not tell you.
Format: Self-paced with roughly 18 topics of labs and two capstone models; assumes the foundations program or equivalent.
Run this program for your team
Every program can be delivered as a private, tailored cohort for your organization, aligned to your systems, policies, and career frameworks.
Scope a corporate cohortFrequently asked questions
What is the Data Science program?
Master applied machine learning and data analytics across finance and energy. Learn Python, modeling, visualization, and MLOps through hands-on case studies and real datasets.
Who is this program for?
It suits aspiring and practicing data scientists, along with others described on this page.
How is it delivered?
Self-paced with roughly 18 topics of labs and two capstone models; assumes the foundations program or equivalent.
Is there a project or capstone?
Deliver two end-to-end models on real datasets: a credit-risk scoring model using logistic regression, and a portfolio-optimization study in Python using the Sharpe ratio and Monte Carlo simulation, each with clean data preparation, honest evaluation, and a clear write-up of what the model does and does not tell you.
How does this fit the wider journey?
The second core-analytics program and the modeling heart of the track. It builds on the SQL foundation and feeds directly into the two applied finance programs, where its methods are specialized to BFSI use cases.
Can my organization run this as a private cohort?
Yes. Every program can be delivered as a tailored corporate cohort. Contact us to scope it.