AWS Builder Center

Using Amazon SageMaker to Build a Data Analytics Project on AWS

Learn how Amazon SageMaker can be integrated into an end-to-end analytics workflow, from data preparation and exploration to machine learning and business intelligence. This article demonstrates how cloud and predictive analytics skills can strengthen a Data Analyst portfolio.

Introduction

As a Data Analyst, working with real-world data often involves more than just SQL, Excel, and Power BI. Modern analytics projects increasingly use cloud platforms and machine learning services to process large datasets, identify patterns, and generate actionable insights.
One AWS service that can add significant value to a data analytics project is Amazon SageMaker. Although SageMaker is primarily known as a machine learning platform, it also provides useful capabilities for data preparation, exploration, experimentation, model development, and prediction.
In this project, I explored how SageMaker can be integrated into a data analytics workflow to build an end-to-end analytics solution on AWS.

Project Objective

The goal of the project was to analyze a business dataset and build a pipeline that could:
  • Store data in AWS
  • Clean and prepare the dataset
  • Explore important business patterns
  • Perform feature engineering
  • Build a machine learning model
  • Evaluate model performance
  • Generate predictions
  • Present insights for business decision-making
The project demonstrates how a Data Analyst can gradually move from traditional analytics toward cloud-based analytics and machine learning.

AWS Architecture

The project workflow was designed around several AWS services:
Amazon S3 → Amazon SageMaker → Amazon S3 → Power BI / Dashboard

Amazon S3

Amazon S3 was used as the central storage layer for the project.
Raw datasets were uploaded to S3 and organized into different folders such as:
project-data/
│
├── raw/
├── processed/
├── features/
└── predictions/

This separation made the data pipeline easier to manage and maintain.

Working with SageMaker

Amazon SageMaker provided the environment for data exploration and machine learning experimentation.
Using SageMaker notebooks, I could access the dataset stored in S3, perform data cleaning, analyze distributions, identify missing values, and prepare the data for modeling.
A typical workflow looked like:
Raw Data
↓
S3
↓
SageMaker
↓
Data Cleaning
↓
Exploratory Data Analysis
↓
Feature Engineering
↓
Machine Learning Model
↓
Predictions
↓
Business Dashboard


Data Cleaning and Preparation

Before building a model, the dataset needed to be prepared.
The major data-cleaning steps included:
  1. Identifying missing values
  1. Removing duplicate records
  1. Correcting incorrect data types
  1. Handling outliers
  1. Encoding categorical variables
  1. Creating useful derived features
  1. Splitting the dataset into training and testing datasets
For example, customer transaction data could be transformed into features such as:
  • Total purchase amount
  • Number of transactions
  • Average order value
  • Customer activity frequency
  • Recency of purchase
These features can provide much more useful information than the raw transaction records.

Exploratory Data Analysis

SageMaker notebooks can also be used for exploratory data analysis.
Using Python libraries such as Pandas, NumPy, Matplotlib, and Seaborn, I analyzed:
  • Data distributions
  • Correlations
  • Trends
  • Outliers
  • Category-level performance
  • Customer behavior
For example, a business might want to answer:
Which customer segment generates the highest revenue?
or:
Which products have high sales but low profitability?
These questions can be answered using a combination of SQL-style analysis, Python-based exploration, and visualization.

Building a Machine Learning Model

After preparing the data, SageMaker can be used to train a machine learning model.
For example, if the objective is to predict customer churn, the workflow could be:
Customer Data
↓
Feature Engineering
↓
Training Dataset
↓
SageMaker Model
↓
Churn Prediction

The model could predict whether a customer is likely to leave.
The output might look like:
Customer
Churn Probability
Customer A
0.82
Customer B
0.21
Customer C
0.67
The Data Analyst can then use these predictions to identify high-risk customers and provide actionable recommendations to the business team.

Model Evaluation

A machine learning model should not be evaluated only by whether it produces predictions.
Depending on the problem, appropriate metrics can be used.
For classification:
  • Accuracy
  • Precision
  • Recall
  • F1-score
  • ROC-AUC
For regression:
  • MAE
  • MSE
  • RMSE
  • R²
For example, if predicting sales revenue, RMSE can help measure how far the predicted values are from the actual values.

Connecting Analytics With Business Intelligence

The most important part of the project is not simply building a machine learning model.
The real value comes from converting the results into business insights.
Predictions generated by SageMaker can be stored back in Amazon S3 and then used as a source for a business intelligence dashboard.
A Power BI dashboard could contain:
  • Revenue KPIs
  • Customer segments
  • Product performance
  • Predicted churn
  • Sales trends
  • Regional performance
  • High-risk customers
This creates a connection between machine learning and business analytics.

Why SageMaker Is Useful for a Data Analyst

A Data Analyst does not necessarily need to become a machine learning engineer to benefit from SageMaker.
SageMaker helps analysts understand how cloud-based analytics pipelines work and how predictive models can support business decisions.
It provides exposure to important concepts such as:
  • Cloud data storage
  • Data preparation
  • Feature engineering
  • Model training
  • Model evaluation
  • Prediction pipelines
  • Data-driven decision making
These skills can be especially valuable for modern Data Analyst and Analytics Engineer roles.

Key Learning Outcomes

Through this project, I learned that analytics is not limited to creating dashboards.
A modern analytics workflow can combine:
SQL + Python + AWS + Machine Learning + Power BI
SQL can be used to extract and transform data.
Python can be used for deeper analysis and experimentation.
AWS can provide scalable cloud infrastructure.
SageMaker can support machine learning workflows.
Power BI can communicate the final insights to business stakeholders.

Conclusion

Amazon SageMaker can be a powerful addition to a Data Analyst portfolio when it is used to solve a real business problem rather than simply demonstrating a machine learning model.
The combination of Amazon S3, SageMaker, Python, SQL, and Power BI provides an excellent foundation for building an end-to-end analytics project on AWS.
The key lesson from this project is simple:
A strong Data Analyst does not just analyze what happened—they use data and predictive techniques to help understand what could happen next and what the business should do about it.
This project demonstrates that transition from traditional reporting to cloud-based, predictive analytics, which is increasingly relevant for modern data roles.
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article