Application of Exploratory Data Analysis to Investigate Factors Affecting Changeover Time of Manufacturing Processes
The manufacturing plant of a leading company in the advertising industry located in Los Angeles faces a long changeover time problem in its manufacturing process. Exploratory Data Analysis (EDA) is a way to summarize a dataset through visualization which is used to confront the long changeover time problem. Published literature to date regarding EDA has indicated that the visualization of data reveals normal as well as unexpected patterns requiring immediate attention. Building on the existing applications of EDA, this study focuses on determining how EDA methods can find out the factors that are affecting the production change over time in manufacturing plants. Based on the literature of EDA, methods of plotting Box Plot, Scatter Plot, Histogram, Density Estimation plot, and Correlation Matrix are used in this study. These EDA techniques are applied to analyze the data using the Python scripting language code in a web application named Jupyter Notebook. The code is written using libraries like pandas, seaborn to create a visual summary of the dataset. Evaluation of the visual summary demonstrated that insert ratio (number of paper inserts in a bunch of advertisements) of a work order is a contributing factor in longer changeover time. Moreover, factors such as Count, and Job Type also show a strong correlation with the changeover time. The data analysis created a theory that factors such as Insert Ratio, Count, and Job Type which contribute to the increase in changeover time. After an investigation conducted in the manufacturing plant in Los Angeles, it was found out that the Insert Ratio is a significant factor contributing to an increase in changeover time. This led to the creation of a new production scheduling application in MS Excel utilizing VBA programming. The Insert Ratio is considered as one of the deciding factors while creating the production schedule. Based on the data analysis and data visualization performed in Jupyter Notebook using Python language, it is concluded that EDA methodologies reveal unexpected factors such as Insert Ratio, Count, and Job Type which contribute to the increase in changeover time.