Predictive analytics and business forecasting- time series regression
Predictive analytics and business forecasting- time series regression.
You will have to select a business problem that interests you (the instructor will suggest a problem in case you run out of ideas.) Problem data/facts should be real, obtained from verified sources, and correspond to realistic situations. Your team will prepare a PowerPoint presentation, to be presented in class. Your team may also submit a written section that includes your SAS code or any coding for other analytical methodology used in this project. The presentation should be no more than 30 slides.
It will utilize time-series analysis to construct a forecast model using SAS. You are expected to collect real-world data, and have it approved before you begin your analysis. The purpose is ultimately to make a predictive model that accurately forecasts a given variable from other values.
Sell – Tell –Compel Method
You should use the following structure to present the findings from your projects. This is the most common method for business/analytics presentations. It will help you organize your project in an easily presentable manner.
Sell your audience on the importance of your research.
The first section is an introduction to the problem that you will be solving. Provide background and context for your presentation. This is also where you will establish the theory for the correlation of variables in your model. Include in this section approx. 5 academic and/or journalistic publications that outline both the need for this predictive ability and why you decided to use the method that you did. (E.g. This article says that heart disease is preventable yet kills thousands every year; these articles say that these factors influence heart disease; therefore, controlling these factors that correlate with heart disease will help prevent it)
Tell the story of your scientific process/research.
The next section includes the steps that you took to reach your conclusion. Start by describing your data, then identify the variables that you used. Then state the hypothesis (H0 and HA). Explain the model that you are using and how you came to this. Show not only your final model, but each model that you tried and why you included, transformed or excluded variables (multicollinearity, norms violated, high p-values, etc.). Please validate your model with residual analysis and/or collinearity (VIF) table. Finally, show the model statistics in table form and any graphs of your model. This section is the heart of your presentation and should represent the majority of your slides.
Compel your audience to action.
The final section is the presentation of your conclusion and implications for the future. If your null hypothesis is rejected, explain what impact this has on the subject of your research. If the null hypothesis is not rejected, then suggest variables to add or remove from your model or how would you modify your theory in future research.
Topics to consider:
· Health and healthcare
· Humanitarian operations
· Disaster relief
· Social services
· Sharing economy
· Urban planning
· Fraud, collusion, and corruption
· Government policy
· Cyber security
· Crime and terrorism
Recommended Journals to use for forecasting:
· International Journal of Forecasting
· Forecasting (MDPI)
· Journal of Forecasting