KWKEISHA WALKER

AVAILABLE FOR REMOTE CONTRACTS

Turning data into
clear decisions.

Data science, machine learning, and business analytics with a practical focus: ask better questions, evaluate the evidence, and make the next step clear.

Explore my work View GitHub ↗
KEISHA WALKERCOLORADO, USA · REMOTE U.S.
THE TOOLKIT

Python · SQL · pandas · scikit-learn · Tableau · Jupyter

01 / SELECTED WORK

Business questions.
Evidence you can explore.

Two featured projects. Choose either project below to explore its case study, methods, and recommendations.

01 / MACHINE LEARNINGMarketing Conversion PredictionPython · scikit-learn · Model evaluationExplore project ↓02 / BUSINESS ANALYTICSNova Retail E-commerce AnalyticsPython · SQL · TableauExplore project ↓
01MACHINE LEARNING · MARKETING

Marketing Conversion Prediction

Who is likely to subscribe?

A classification project examining bank campaign subscriptions—and what happens when customer response patterns change over time.

Pythonscikit-learnClassification
  • Logistic regression & random forest comparison
  • Chronological evaluation & leakage prevention
  • Model interpretation & business recommendations
Read case study ↗Explore code ↗
Evaluation notes

The final evaluation showed limited discrimination (ROC AUC 0.538). Subscription rates changed substantially across chronological splits. The final holdout block had been examined earlier; it is not a new independent test. These findings inform further validation before operational use.

A CLOSER LOOK / OUTCOME DRIFT
Observed subscription rates increase across chronological splits: approximately 4.81% training, 11.07% validation, and 30.83% test.
41,188
records in the UCI Bank Marketing dataset
Real public data · documented source

A model’s context matters as much as its score.

NOVA RETAIL / BUSINESS ANALYTICS

The story
behind the sale.

01DATA QUALITY
02SALES & PROFIT
03MARKETING
04RECOMMENDATIONS

Synthetic retail data · Business case study

02E-COMMERCE · BUSINESS INTELLIGENCE

Nova Retail
E-commerce Analytics

Nova Retail connects data quality, sales, profitability, and marketing analysis to practical business recommendations through a Python, SQL, and Tableau workflow.

PythonSQLTableau

Published materials include the business case study and original synthetic source-data package.

Read case study ↗View repository ↗
Project materials

This case study uses a fictional business and synthetic data. The repository currently includes the report and original data; finished analysis scripts and the Tableau workbook are not yet included.

02 / FOCUS AREAS

Technical thinking.
Practical application.

Data science at the center. Business context throughout.

01 / PREDICT

Predictive analytics

Classification, baseline comparisons, model evaluation, and interpretation tied to a clearly defined business question.

02 / UNDERSTAND

Business intelligence

SQL analysis, data quality assessment, visualization, and concise reporting that makes findings easier to act on.

03 / AUTOMATE

AI & agent workflows

Applied AI interests include workflow automation, business-document assistants, and agents that connect tools and APIs—with evaluation and human review.

03 / ABOUT

Hi, I’m Keisha.
I connect the details
to the bigger picture.

I use Python, SQL, and data visualization to investigate business questions, evaluate predictive models, and translate analysis into practical recommendations.

My approach is straightforward: understand the problem, validate the data, compare against a baseline, and explain the findings clearly. Strong work includes documented assumptions and a realistic account of what the evidence supports.

BASED INColorado, USA
EDUCATIONBachelor’s in Data Science
In progress

04 / LET’S CONNECT

A good project starts
with a clear question.

Available for remote freelance, independent-contractor, and 1099 opportunities in data science, analytics, model evaluation, and AI automation.

Discuss a project ↗

Keishabonit@gmail.com

Share your business question, project scope, and timeline. View GitHub ↗