DATA4400 Data Driven Decision Making and Forecasting Report 2 Sample

DATA4400 Data Driven Decision Making and Forecasting

Assessment Details

Your Task

Given a dataset with multivariate time series data, you are to conduct multiple forecasting methods and provide a description and interpretation of the techniques used.

A dataset will be provided to you at the beginning of week 9. The objective of the assessment is to build different forecasting models using Orange Data Mining and Tableau. Students must calculate the Root Mean Square Error (RMSE) or Mean Absolute Percentage Error (MAPE) to evaluate the performance and accuracy of the model, as well as choose the appropriate metrics for model selection.

Report Structure and Content

Imagine you work for the Central Bank of Genovia and your task is to forecast the unemployment rate in one quarter.

1. Import the DATA4400_A2_Data.csv dataset into Orange Data Mining (https://orangedatamining.com/).

2. Assess the quality of the data in terms of missing values and provide summary statistics of the variables.

3. Using an ARIMA model, forecast the unemployment rate for one quarter.

a. What is the forecast unemployment rate based on the ARIMA model?

b. Provide a screenshot of the ARIMA model settings and the appropriate visualisation for your forecast.

4. Using a VAR model, forecast the unemployment rate for one quarter.

a. What is the forecast unemployment rate based on the VAR model?

b. Provide a screenshot of the VAR model settings and the appropriate visualisation for your forecast.

5. How do the Fed Funds rate and the unemployment rate affect each other in Genovia?

6. Use Tableau (https://www.tableau.com/academic/students) to visualise the dataset and generate a forecast of the unemployment rate at the end of the next quarter.

7. What is the unemployment rate forecasted by Tableau?

8. Explain which model was used in Tableau and report on its parameters.

9. Evaluate the models using the available metrics and report which model provides the best forecast.

10.Summary

Solution

Introduction

The Central Bank of Genovia must evaluate forecasting-based analytics in the dynamic sector of financial analysis, according to the paper. This study maximizes Orange Data Mining using Tableau to forecast the unemployment rate for the quarter (Ranathunga, 2022). Methodically calculating performance indicators like MAPE and RMSE to develop multiple forecasting models is the goal. They guide precision and accuracy and help choose the optimal forecasting model. This volatile climate makes foresight crucial to Genovia's financial decisions.

Analysis and Implementation

In order to reduce error and improve prediction accuracy for the projected unemployment rate, they have examined and put into practice a variety of forecasting models using Orange Data Mining and Tableau (Coelho et al. 2022).

Orange Data Mining

Dataset Import:

Figure 1: Dataset Importing
(Source: self-created in Orange Data Mining )

After importing the "DATA4400_A2_Data.csv" dataset into Orange Data Mining, it was found that there was not a single instance of missing data, which brought the total number of instances to 756, in total. The dataset is comprised of a total of five unique features, which are denoted as "Date," "CPIAUCSL," "INDPRO," "FedFundsRate," and "UnemploymentRate." Notably, neither the intended target variable nor any anticipated meta-characteristics were discovered inside this dataset (Hasan et al. 2020).

Data Preprocessing:

Figure 2: Feature Statistics
(Source: self-created in Orange Data Mining )

Feature statistics is one of the useful statistical concepts that is used in this study. The dataset is explored by using the feature statistics in this study. The orange data mining process is applied in this study to calculate the mean, median, mode, dispersion, min, max, and milling values of the unemployment rate, INDPRO, CPIAUCSL, FedFundsRate, and date. The mean value of the unemployment rate has been generated which is 5.960. The mode value of the unemployment rate is 5.4. The median value of the unemployment rate is 5.7. The dispersion and min. The values of the unemployment rate are 0.234 and 3.4. The max. and missing values of the unemployment rate are 10.8 and 0(0%). In Orange Data Mining thoroughly evaluates data quality, focusing on missing values and variable summary statistics. Pretreatment provides the dataset for detailed and insightful analysis, promoting a thorough understanding of its qualities and improving forecasting models (Banimustafa and Hardy, 2020).

Model Implementation:

ARIMA mode:

Orange Data Mining makes it easy to forecast quarterly unemployment rates using an ARIMA model (Schmidl et al. 2022). This accurate time-series forecasting method combines moving average (MA), autoregressive (AR), and differencing components with other methods. After thorough data preparation, the ARIMA model was run to estimate future unemployment rates. The ARIMA model's predicted unemployment rate helps quantify the economy's next quarter.

a. Forecast unemployment rate:

Figure 3: ARIMA Model and Evaluation
(Source: self-created in Orange Data Mining )

The above ARIMA model is developed for forecasting the unemployment rate. The mane of the ERIMA model which is used in this research is ARIMA(1,2,1). The auto-regression order range of the above ARIMA model is 1 and the Differencing degree is 2. The moving average order is 1 and the forecast step ahead is 6. The number of floods is 20 and the forecast steps are 3 have been set to evaluate the parameters of model evaluation.A number of ARIMA model implementation parameters were carefully chosen to optimize forecasting accuracy. The order of differencing (d), autoregressive order (p), and moving average order (q) were adjusted to fit the dataset. To make expected outcomes easier to read, appropriate for MBA assignment expert visualizations were generated. A screenshot of the ARIMA model settings shows the parameter values and visualizations like time-series plots and predicted vs. actual values.

b. Data visualization:

Figure 4: Line Chart
(Source: self-created in Orange Data Mining )

Orange Data Mining uses historical unemployment rate data and the ARIMA algorithm to help users make informed decisions and strategic strategies. Using the Line Chart the information on the unemployment rate is shown which has been changed over time. The above Line chart represents the unemployment rate between 1955 to 2017. The unemployment rate increasing pattern is clearly visible in the graph. While there were times of high growth between 1980 and 1990, and between 1960 to 1970 and again between 1990 and 2010, there were also periods of poor growth between 1960 and 2010.

VAR model:

An efficient Vector Autoregression (VAR) model implementation in Orange Data Mining predicts the unemployment rate for a future quarter. The VAR model is appropriate for capturing complex economic data linkages because it accounts for time-series variable interdependencies (Ji et al. 2019). After careful data preparation and VAR model parameter setting, it accurately predicted unemployment.

a. Forecast unemployment rate:

Figure 5: VAR Model and Evaluation
(Source: self-created in Orange Data Mining )

The predicted unemployment rate, which was obtained from the complex analysis of historical data by the VAR model, offers an invaluable quantitative forecast of the state of the economy. With careful consideration given to the lag order selection, variable inclusion, and other pertinent criteria, the VAR model settings were fine-tuned to produce the best results. In order to make the predicted outcomes easier to understand, engaging visuals were also created. The VAR model parameters, selected parameter values, and a useful visualization, such as a time-series plot showing projected unemployment rates, have been intelligently developed and are shown in a screenshot.

b. Data visualization:

Figure 6: Line Chart
(Source: self-created in Orange Data Mining )

the VAR model's effectiveness resides in its capacity to identify complex correlations between many economic variables, facilitating a comprehensive comprehension of economic processes. Using the Line Chart the information on the unemployment rate is shown which has changed over time. The above Line chart represents the unemployment rate between 1955 to 2017. The unemployment rate increasing pattern is clearly visible in the graph. While there were times of high growth after 1980, between 1960 to 2010, there were also periods of poor growth between 1960 and 2000.

Unemployment rates affect each other in Genovia

The connection between the Fed Funds rate and the unemployment rate in Genovia has been noted to be complex. In the past, lower Fed Funds rates have been used to boost economic expansion and lower unemployment. In contrast, higher rates are used to control inflation but may unintentionally lead to higher unemployment.

Tableau

Visualise the dataset and generate a forecast of the unemployment rate

Figure 7: Forecast of the unemployment rate at the end of the next quarter
(Source: self-created in Tableau )

The dataset has been efficiently represented with Tableau, providing perceptive data representations (Qin et al. 2020). An estimate of the unemployment rate for the end of the following quarter has also been carefully created.

Figure 8: Unemployment rate
(Source: self-created in Tableau )

The unemployment rate is predicted to be 66.07% in 2018 by the forecast indicator "Estimate" of Tableau. The Central Bank of Genovia uses this predicted rate as a crucial quantitative indicator to help with economic planning and decision-making.

Figure 9: Forecast Trend Model
(Source: self-created in Tableau )

Forecasting in Tableau uses the "Trend Lines Model." Based on the Year of Date, this model creates a linear trend for the sum of actual and anticipated unemployment rates. The significance of the model (p <= 0.05) indicates its reliability. "Forecast indicator * (Year of Date + intercept)," with 370 modelled observations and no filters, is utilized. The model has 4 degrees of freedom and showed strong significance (p-value < 0.0001) in the analysis of variance. Coefficients and related statistics, such as t-values and p-values, help explain the forecasting process by revealing model parameters.

Best Forecast Evaluate the models using the available metrics

The model that provides the best forecast, according to an evaluation of the given data, is the "Trend Lines Model." Based on the selected evaluation criteria, this model performs better than others, showcasing its higher predicting accuracy within the specified context.

Conclusion

The analysis of the data demonstrates that there have been considerable shifts in the unemployment rate in Genovia over the course of time. These patterns shed light on the importance of adopting flexible economic policies in order to respond to times of rapid economic expansion and maintain historically low Unemployment rates.

Reference List

Ranathunga, P., 2022. Machine learning based sales forecasting system (Doctoral dissertation, Robert Gordon University, Aberdeen, UK 2022).

Hasan, R., Palaniappan, S., Mahmood, S., Abbas, A., Sarker, K.U. and Sattar, M.U., 2020. Predicting student performance in higher educational institutions using video learning analytics and data mining techniques. Applied Sciences, 10(11), p.3894.

Banimustafa, A. and Hardy, N., 2020. A scientific knowledge discovery and data mining process model for metabolomics. Ieee Access, 8, pp.209964-210005.

Schmidl, S., Wenig, P. and Papenbrock, T., 2022. Anomaly detection in time series: a comprehensive evaluation. Proceedings of the VLDB Endowment, 15(9), pp.1779-1797.
Ji, Q., Bouri, E., Roubaud, D. and Kristoufek, L., 2019. Information interdependence among energy, cryptocurrency and major commodity markets. Energy Economics, 81, pp.1042-1055.

Qin, X., Luo, Y., Tang, N. and Li, G., 2020. Making data visualization more efficient and effective: a survey. The VLDB Journal, 29, pp.93-117.
Coelho, D., Gupta, N., Papenhausen, E. and Mueller, K., 2022, November. Patterns of Social Vulnerability-An Interactive Dashboard to Explore Risks to Public Health on the US County Level. In 2022 Workshop on Visual Analytics in Healthcare (VAHC) (pp. 01-09). IEEE.