Data Analysis

Core Services 

Data Cleaning & Preparation

Turning chaotic spreadsheets and databases into clean, reliable datasets.

Exploratory Data Analysis (EDA)

Deep statistical insights with beautiful, interactive visualizations.

Predictive Modeling & Machine Learning

Forecasting, churn prediction, classification, clustering, and more.

Interactive Dashboards & Reporting

Custom Streamlit or Plotly Dash apps that update automatically.

Business Intelligence Automation

Replace slow Excel processes with scalable Python pipelines.

Data Strategy Consulting

One-time audits or ongoing support to build your data capabilities.

Featured Projects

Production Efficiency & Predictive Maintenance Dashboard

Client: Plastics Manufacturer in Gauteng, South Africa

Challenge: High machine downtime and 11% scrap rate due to poor visibility in production data.

Solution: Built a Python-powered predictive maintenance system with real-time dashboards. Used Pandas/Polars for data processing, XGBoost for failure prediction, and Streamlit for an interactive web app.

Results:

  • OEE improved from 62% to 78%
  • Unplanned downtime reduced by 41%
  • Scrap rate dropped to 4.8%, saving ~R1.4 million annually

Tools: Python • Pandas • XGBoost • Plotly • Streamlit • PostgreSQL

Loan Default Risk Analysis & Dashboard

Client: A South African bank

Challenge: High bad debt levels due to inaccurate manual risk assessment and lack of predictive insights in the loan approval process.

Solution: Developed a machine learning model using Python to predict loan defaults. Created an interactive dashboard that displays risk scores, key influencing factors, and automated weekly reports for the credit team.

Results:

  • Reduced bad debt exposure by 24%

  • Improved default prediction accuracy significantly

  • Enabled faster and more confident lending decisions

Tools: Python • Pandas • Scikit-learn • Seaborn • Streamlit

Testimonials

“David transformed our chaotic production data into a powerful decision-making tool. The predictive dashboard has paid for itself many times over. Highly recommended for any manufacturer looking to modernise with data.”

James L., Operations Director

“David delivered an outstanding loan default prediction model and interactive dashboard for our personal loans division. His Python-based solution (using Scikit-learn and Streamlit) gave us clear risk scores and driver insights that we previously lacked. Within months, we reduced bad debt exposure by 24%. He was professional, fast, and explained complex models in simple business terms. Highly recommended for any bank or fintech needing reliable data solutions.”

Karen W., Senior Credit Risk Manager

Let's communicate

Address:

Pretoria, GP

South Africa (ZA)

© David Nchabeleng

2026