Assignment Remit

Assignment Remit

This assignment involves a practical task. You are asked to source a set of data, clean , manipulate, and use it to produce insights that would be useful to a specific audience or for a defined purpose. You need to produce a report of the process undertaken and the tools you have used for collecting, processing, and analysing data. For this assignment, each student is required to:

1.   Select an Appropriate Dataset. Begin by carefully selecting a dataset to apply for this assignment. Your selection can be from any sources so long as there are no copyright restrictions that limit the use of the data. The dataset(s) you select should be those that you think are interesting to a particular group of people. You can find your own sources and take some guidance from lecture material. Other sources of useful datasets might be:

-  Google Dataset Search -https://datasetsearch.research.google.com

- Data.gov.uk

-  https://data.gov

-  UK Data Service - https://ukdataservice.ac.uk

-  World Bank DataBank - https://databank.worldbank.org

-  OECD Data Explorer -https://data-explorer.oecd.org/

-  Eurostat Database -https://ec.europa.eu/eurostat/web/main/data/database

-  Etc.

You can scrape data from public websites where this is appropriate - but we are looking for rather large datasets (more than 500 observations/records with more than 6-7 variables/features) as a key element of your Data Story—not just a few numbers.

NB- Avoid redundancy by ensuring that the chosen dataset differs from the one utilized in your work group assignment. In other words, refrain from using the same dataset for both Individual and Group Work Assignments.

2.   Identify the Target Audience. Clearly define the target audience for your

visualizations. Explain why this audience would be interested in the data and how they are expected to use it once provided.

3.   Prepare the Dataset. The dataset you find may need restructuring, cleaning and editing to improve its quality and suitability for your purpose. You may use any tool to clean the data, including (but not necessarily limited to) Python, Excel or  the data cleaning tools embedded in Tableau software.

4.   Perform Exploratory Data Analysis (EDA) to understand and summarize the key characteristics of a dataset, identify patterns or anomalies, and prepare the data for modelling. Use Excel or Tableau (or both) to explore the data. Your EDA should include a range of visualizations, such as:

• Histograms to examine distributions of numeric variables.

• Boxplots to detect outliers and understand the spread of data.

• Scatterplots to explore relationships between pairs of variables.

• Bar charts, line charts, or PivotCharts to analyze trends or compare categories.

• Summary tables to present counts or percentages of each categorical

variables, and descriptive statistics like mean, median, mode, and standard deviation for numerical variables.

• Other relevant visual tools based on the characteristics of your dataset. The primary objectives of your EDA are to:

• Gain a good understanding of the variables, including the main characteristics of each variable (e.g., distributions, central tendencies, and variability).

• Identify patterns, trends, and relationships between variables.

• Detect missing values, outliers, or anomalies in the data and propose strategies to handle them.

5.   Select and build appropriate data modelling. The choice of model depends on the nature of the business problem , the goals of the analysis, and the structure of your dataset. This may involve:

Regression model:

•     Purpose: Analyse and predict numerical dependent/outcome variable based on independent variables/predictors.

•     Examples: Predicting sales revenue, customer lifetime value, or housing prices.

•    Approach:

o  Select appropriate regression techniques such as multiple linear regression or non-linear regression models

o  Evaluate the model's performance using metrics like R-squared and interpret coefficients.

Times-series model:

•     Purpose: Understand and model patterns, trends, and temporal dependencies in time-ordered data.

•     Examples: Forecasting sales, stock prices, demands, or website traffic trends.

•    Approach:

o  Decompose the time series into trend, seasonality, and residual components.

o  Apply models like Moving Average, Exponential Smoothing, and Regression-based forecasting.

o  Evaluate predictions using metrics like Mean squared error (MSE) or Mean Absolute Percentage Error (MAPE).

Classification model:

•     Purpose: Analyse and predict categorical dependent/outcome variable based on independent variables/predictors.

•     Examples: Fraud detection, customer risk classification, or churn behaviour.

•    Approach:

o  Use classification algorithms such as k-Nearest Neighbours, logistic regression or decision trees.

o  Evaluate model performance with metrics like accuracy.

Unsupervised model:

•     Purpose: Identify natural groupings or patterns within the data without predefined labels.

•     Examples: Market/Customer segmentation, grouping similar products, market-basket analysis or text mining.

•    Approach:

o  Use unsupervised algorithms like K-Means, hierarchical clustering, sentiment analysis.

o  Evaluate cluster validity using metrics such as silhouette score or Elbow method

o  Visualize clusters using scatter plots, dendrograms, or silhouette plots.

Structure of your data analytics report (2,500 words maximum):

Chapter 1- Business Understanding:

•    Detail who the target audience is and the purpose for which they might use the data analytics results.

Chapter 2- Data Understanding:

•    Explain why you chose the dataset(s) you did. You must provide a link to the dataset(s) used.

•    Describe the data: its size in terms of number of records (observations) and variables (features). Provide data dictionary, including the variable names, formats (e.g., numeric, categorical), descriptions, and examples of data values.

Note: Ensure the dataset is different from the one used in your group assignment to avoid redundancy.

Chapter 3- Data Preparation:

•    Explain the process you used to clean, edit or constructing the data. If you discarded any data say why this was done.

•    If you merged or integrate datasets, explain how and why you did this.

•    Describe what problems you encountered and how you overcame them.

Chapter 4- Exploratory Data Analysis:

•    Use the methods of descriptive statistics and visualisation (such as

crosstabulation, histogram, bar charts, line charts, scatterplots, heat maps, PivotCharts, etc) to explore the data.

•    Explain and interpret the results of the visualisations, highlighting trends, patterns, relationships, or anomalies. Discuss the implications of these findings in the context of the business or problem at hand.

Chapter 5- Modelling:

•    Depends on the business problem and dataset, apply regression, times   series analysis, classification or clustering techniques to build business analytics model(s).

•    Explain and interpret the results of the model(s) in relation to the business objectives.

•    If applicable, include comparisons of multiple models to identify the best- performing approach.

Submission guidance:

1.   Ensure each chapter flows logically and builds upon the previous one and keep

the report concise, focusing on key insights and actionable findings.

2.   Include the screengrabs of data visualisations in your word submission to help give the word document context.

3.   Provide links to the source dataset(s) you used - otherwise we cannot audit the validity of your data, and you will drop marks.

4.  Your report should be submitted in the form. of a Word document, use minimum 12pt font, and at least 1.2 line spacing.

Module Learning Outcomes:

This assignment tests the following module learning outcomes:

•    Collect, analyse and interpret data analytics to make informed business decisions.

•    Appraise how digital business and data analytics can be used to generate actionable insights for managers and decision-makers.

•    Communicating, presenting and disseminating analysis of the data.