Reference no: EM133818119

Data-driven Forecasting

Assessment - Forecasting for a Business Problem

Your Task
Develop a real-world forecasting project plan/proposal based on the learnings from the subject. This assessment is to be done individually.

Part A - 1.5 weeks before the report is due, in Week 9, students are to record a 5-minute video talking about what progress they have made so far in their report with evidence to support.

Part B - Students are to write a 1000-word report on Forecasting for a Business Problem.

Assessment Description
This assessment seeks to simulate a real-world task that you may have to undertake in the future. Therefore, the assignment is non-prescriptive and requires you to pose a relevant, small, creative and significant forecasting problem to solve that could result in benefits to the organisation of choice.

In this assessment, you need to consider an organisation in an industry of your choice and articulate the steps this organisation needs to take to enable forecasting for data-driven decision making. You are required to source an example time series, or maybe more than one time series, to demonstrate expected forecasting outcomes.

The application does not need to be of major financial significance to the general community, for example, for one student it was the number of people viewing her YouTube channel. Another was a waste of fresh fruit and vegetables at a small grocery store. A small well-focused example is ideal.

The analysis can be quite simple and performed in Excel, if appropriate. Aim for an interesting and original application of time series forecasting. You are not expected to compare different forecasting methods or to split time series into train and test segments.

Plan A
Ideally, the application is based on an organisation you work for, or have worked for, in Australia or overseas. Use genuine time series data - though it can be scaled/perturbed if commercial confidentiality is an issue. A short time series of annual data may suffice. For example, a student was working for a not-for-profit organisation that runs care homes - annual data on several clients was read from annual reports from the past 8 years. Another student was working in a small grocery store. There was considerable wastage of fresh fruit and vegetables, but the store would not provide data. One approach would be to simulate realistic demand, based on your observations, and then investigate how a strategy based on Holt-Winters forecasts would reduce waste. Your facilitator can advise on the feasibility and suitability of using simulated time series for a specific application.

Plan B
Alternatively, you'll need to source some genuine up-to-date time series data and imagine some scenarios. For example, you might imagine you run a car dealership and need to order a mix of petrol and electric vehicles. Look for the time series of EV sales and consider the relationship with the number of charging facilities. Some interesting data sets can be found on GovHack - one of the 2024 challenges was to forecast tourist numbers for the Northern Territory. Other possibilities are public health issues including road traffic accidents. When you write up your application you can imagine you are working in a relevant organisation.

Plan C
Share prices are readily available and can be downloaded from Yahoo. Give some context. For example, imagine you are a stockbroker, and you have a client who wishes to invest in one of the four Australian clearing banks (ANZ, CBA, NAB, WESTPAC). Which one would out recommend for a one-year investment? You work for a company that offers an Employee Share Scheme - is this a good investment for you? You might investigate Granger causality between share prices, between a share price and volume traded, or between a share price and an economic indicator such as base rate, unemployment, building approvals etc. Remember that the VAR model used for assessing Granger causality is stationary - so work with log-returns of share prices. Hire Best Homework Experts Now!

Sample datasets maybe sourced from:
an organisation if you work there,
public repositories
Open government data
Open data such as share prices and commodity prices.
Please do not use Kaggle data- mostly

Could you attempt Plan A? If so, how did you progress? Do you have relevant data? If you are now going with Plan B, or C tell us about the application.

Record a 5-minute video, including slides and a live demonstration of your accessing the timeseries and plotting it, that covers the following points.

The company of choice and the business problem identified. [1 or 2 slides]
How forecasting will be a solution to this problem. [1 or 2 slides]
What data you have gathered, including the source and unit of measurement. [1 or 2 slides]
Demonstrate loading the timeseries you have sourced into any convenient software and then demonstrate plotting the timeseries. You can use different software for the report, if you wish to do so. [live demonstration]
What progress you have made so far including pain points/hurdles/issues/challenges you are facing with your written report. [live discussion]
You need to Share Screen to record the slides, then Stop Share to record the live demonstration and discussion.

Part B - The report should address:
The organisation you have chosen and your familiarity with its business.
Why forecasting would help this organisation given their current operations.
The time series you have sourced. The source of the time series. Details of the variable(s) including unit of measurement.
The forecasting technique you would recommend, with your reasons for doing so. If you have more than one time series you can consider VAR models or Prophet with External Predictors. You are NOT EXPECTED TO fit other forecasting models and compare error statistics, or to compare training and test error statistics. You are expected interpret your results in a meaningful way. You can use a software of your choice for the methodology/analysis.
The results of applying your chosen technique to your time series, including visual displays, the actual forecasts, and error metrics to comment on accuracy. Provide helpful captions for all figures and tables.
The benefits of this forecasting project for the organisation (The benefit could be financial, such as Return on Investment (ROI) or societal benefits).

Assessment Instructions

By the beginning of Week 7 Choose from either Plan A, Plan B, or Plan C. Identify a company and industry you are familiar with that would benefit from a forecasting application. Note:
The application needs to be based on forecasting (not some other aspect of analytics).
Focus on a single, well-defined (small) application.

By the beginning of Week 9 draft some preliminary points pertaining to the report. You are encouraged to consider the current mode of operation, possible inefficiencies, available data, and how this data may be used to provide efficiencies based on the concepts and techniques covered in the subject. Think of yourself as a consultant or a founder.

Your facilitator will advise on the appropriateness of your choice and proposed methodology regarding the requirements for the assessment.

Include a list of references that are directly related to the content. Each reference needs to be linked to at least one specific point in the content of your assessment. It is expected that you will have at least 4 relevant references, web sites are ok, including a reference to the source of the data.