Paper For Above instruction
The aim of this project is to analyze employee transportation preferences with an emphasis on understanding the factors influencing the choice to use a car as a mode of transport. To achieve this, a comprehensive exploratory data analysis (EDA) will be conducted on the dataset, followed by a multicollinearity assessment to refine predictive variables. This report encompasses insights derived from the data, identifies significant predictors, and discusses how multicollinearity is detected and addressed.
Introduction
Understanding employees' commuting modes is vital for organizations aiming to promote sustainable transportation options and improve workforce management. The dataset provided, 'Cars.csv', offers a mixture of personal demographic information and professional background, along with their chosen mode of transportation. The primary focus is to build a predictive understanding of whether an employee is likely to use a car for commuting, based on these features. A detailed exploratory analysis serves as a fundamental step to visualize data distributions, uncover patterns, and identify potential predictor variables. Moreover, multicollinearity among predictors must be examined because high correlations between independent variables can distort the model's interpretation and performance.
Exploratory Data Analysis (EDA)
The first step involves importing the dataset and gaining a preliminary understanding of its structure. This includes inspecting variable types, distributions, and missing values. Descriptive statistics provide insight into the central tendency and variability of numerical features like age, salary, and work experience. Visualizations such as histograms and box plots for continuous variables reveal distribution patterns and
outliers. For categorical variables like mode of transport, bar charts depict frequency counts.
Furthermore, cross-tabulations and group-based visualizations explore relationships between categorical and numerical variables. For example, salary distributions segmented by transport mode may reveal whether higher earners prefer cars. Similarly, correlations between numerical features are examined via correlation matrices. These initial insights guide feature selection and modeling strategies.
Insights Based on EDA
Preliminary analysis suggests that certain variables significantly influence the likelihood of using a car. Typically, higher salaries are associated with increased car usage, possibly due to affordability. Age and work experience may also correlate, with older employees possibly favoring cars for convenience or status.
The distribution of transport modes indicates that a subset of employees predominantly use public transport or other means, providing contrast to car users. Outliers and unusual patterns identified through box plots necessitate data cleaning for accurate modeling.
Multicollinearity Assessment
To detect multicollinearity, a correlation matrix of numerical predictor variables is computed. High correlation coefficients (e.g., above 0.8) suggest multicollinearity. Visual tools such as heatmaps further facilitate this detection. Variables with high collinearity potentially cause instability in regression coefficients, impairing model interpretability. To address this, variables exhibiting multicollinearity are examined further, and some may be removed or combined.
For example, age and work experience are often highly correlated; retaining both may introduce redundancy. Variance Inflation Factor (VIF) analysis quantifies the severity of multicollinearity. Variables with high VIF values (e.g., above 5 or 10) are candidates for removal or transformation. Adjustments improve the robustness and interpretability of subsequent predictive models.
Plotting scatterplots between highly correlated variables helps visualize their relationship, supporting the decision about variable treatment. If necessary, dimensionality reduction techniques like Principal Component Analysis (PCA) can be employed to combine correlated features into a single predictor.
Conclusion
This exploratory phase provides valuable insights into employee transportation choices, highlighting key predictors such as salary and age, while also identifying multicollinearity issues among predictors like age
and work experience. Addressing multicollinearity ensures more stable and interpretable models, ultimately aiding in accurate prediction of employees' mode of transport preferences. Future steps involve model development using these refined predictors to classify whether an employee is likely to use a car, informing organizational strategies for transportation management and sustainability initiatives.
References
Agresti, A. (2018).
Statistical methods for social sciences . Pearson Education.
James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013).
An introduction to statistical learning . Springer.
Montgomery, D. C., & Runger, G. C. (2014).
Applied statistics and probability for engineers . Wiley.
Friendly, M. (2002). Corrgrams: Exploratory data analysis in coast.
The American Statistician , 56(4), 316-324.
Kutner, M. H., Nachtsheim, C. J., Neter, J., & Li, W. (2005).
Applied linear statistical models . McGraw-Hill.
Faraway, J.J. (2014).
Linear models with R . Chapman and Hall/CRC.
Ghazali, N. A., & Ramli, R. (2019). Addressing multicollinearity issues in regression analysis.
International Journal of Academic Research in Business and Social Sciences , 9(7), 434-446.
James, G., et al. (2017).
An introduction to statistical learning with applications in R . Springer.
Everitt, B., & Hothorn, T. (2011).
An introduction to applied multivariate analysis with R . Springer.
Zuur, A. F., Ieno, E. N., & Smith, G. M. (2007).
Analyzing ecological data . Springer.
