Description
MIE1628
Assignment on Azure Cloud Platform
PART A:
1. [Marks: 5] Explain below the 5 components shown in orange boxes. Explain which Azure
components you will use where in this big data architecture and why.
Raw Data
Unstructured Data
Structured Data
Ingest Data Data Store
Prepare and
transform data Model and
serve data
Azure
Databricks
Azure Data
Factory
Azure Synapse
Analytics
Azure Cosmos
DB Azure Data
Lake
Cloud-based Data Analytics
MIE1628
Assignment 5
2. [Marks: 5] Explain how Stream Analytics works in Azure. Mention at least two common use
cases or applications for this service.
3. [Marks: 10] Deploy all the resources in Azure Portal. Implement a Stream Analytics job by
using the Azure portal. See this for reference – https://learn.microsoft.com/en-us/azure/streamanalytics/stream-analytics-quick-create-portal
For query use below:
SELECT *
INTO BlobOutput
FROM IoTHubInput
HAVING Temperature > 25
See the below screenshot and show the top 30 results for your output.
Part B:
Data Input: Claim a dataset from Piazza – link. If the dataset is too large, you can take a subset of the data
as well. No two groups can have the same dataset.
Your selected dataset should meet the following criteria:
1. It must contain a minimum of 1,000 instances (rows or data points).
2. It should include at least six features (columns or attributes).
Using this dataset, you are required to address a substantial and meaningful problem. Your analysis should
demonstrate:
1. A clear understanding of the dataset’s context and potential applications.
2. The ability to formulate relevant questions or hypotheses based on the data.
3. Appropriate use of data analysis techniques to extract insights.
Cloud-based Data Analytics
MIE1628
Assignment 5
4. The capacity to draw meaningful conclusions that could inform decision-making or further
research.
Some problems to consider:
1. Fraud Detection System
2. Customer Churn Rate Prediction
3. Segmentation using Clustering
4. Recommendations with your Dataset
5. Sales Forecasting
6. Stock Price Predictions
7. Human Activity Recognition with Smartphones
8. Wine Quality Predictions
9. Breast Cancer Prediction
10. Sorting of Specific Tweets on Twitter etc.
Implement this part in Azure Machine learning using Azure Notebook
1. [Marks: 15] Clearly define the problem you intend to address using this dataset. Present a
comprehensive problem statement that includes:
a. A detailed description of the meaningful issue you’re tackling
b. An outline of all necessary steps, including:
i. Data preprocessing
ii. Data cleaning
iii. Modeling approach
Your problem statement should be thorough, spanning approximately half to one full page. If you
determine that data cleaning is unnecessary, please provide a justification for why this dataset
doesn’t require cleaning. In such a case, allocate more attention to other crucial aspects such as
EDA and the modeling process.
Ensure your problem statement is well-structured, coherent, and provides a clear roadmap for your
data analysis project.
2. [Marks: 10] Explore your dataset and provide at least 5 meaningful charts/graphs with an
explanation.
3. [Marks: 10] Do data cleaning/pre-processing as required and explain what you have done for your
dataset and why?
4. [Marks: 15] Implement 2 machine learning models and explain which algorithms you have selected
and why. Compare them and show success metrics (Accuracy/RMSE/Confusion Matrix) as per
your problem. Explain results.
5. [Marks: 15] Deploy a run-time pipeline for your dataset using Azure Designer Studio.
Or
Do hyperparameter tuning for your algorithms. Explain your results.
Or
Use Automated ML for your data set. Explain the best model results.
Cloud-based Data Analytics
MIE1628
Assignment 5
6. [Marks: 15] Summarize your project’s key findings and overall conclusions in a brief paragraph.
Ensure your summary is firmly grounded in the data and analysis you’ve presented throughout your
project. Offer meaningful insights that not only encapsulate your work but also lay a foundation
for potential future research in this area. Your conclusions should be well-reasoned and directly
supported by your results.
