Working as a Data Analyst - Process

It’s the end of August, the summer for 2023 has coming to the end. An old saying goes from Kong Zi state “逝者如斯夫，不舍昼夜”, comparing time to the river, which always going forward and going fast no matter day or night. I can still remember the day I stepped into McMaster Manufacturing Research Institute. At that time, some construction on the back was not even finished, but right now, most of them are finished and lots of things has happened.

Being a data analyst in the manufacturing sector, especially as a newcomer to the profession, presents both a sense of challenge and a significant weight of responsibility. In this role, my primary responsibility involves uncovering the latent patterns embedded within the data I gather. Specifically within my context, it's akin to utilizing a stethoscope to audibly discern the inner workings of machinery, attempting to pinpoint the origins of issues within the enormous body of steel.

In Go Game, there is the word “一招不慎，满盘皆输”, which means every step matters, one bad move could cause overall failure. Same to data work, since it is never simply writing the script and preparing a dashboard, it starts with Data collection, data organization, data cleaning, data processing and analysis and finally data visualization.

Throughout the course of my work, I had the opportunity to engage with each of these stages, allowing me to comprehend the difficulties inherent to each.

For example, I got a chance to visit the factory we cooperate with and visualize and assist the DAQ process. I helped to remove the metal scraps and residual engine oil on the surface to place the sensor on. To make the collection process standardized, we need to make sure the place we place the sensor is approximate even though the motor has a different size and location. Additionally, the cord connected between the sensor and the computer needs to be organized and sorted properly to remove any interference effect. Finally, after everything is set up, we need to cooperate with the operator closely to ensure we pick up the signal at approximately the same time. Human error could happen at any second of the DAQ process as any unexpected vibration or interference could contaminate the data. Hence, what we need to do is to keep our process standardized and be careful for every movement we make.

It is imperative to ensure the smooth progression of all these stages. For instance, anomalies, unusual vibrations, or external interferences might infiltrate the dataset due to abrupt movements during DAQ, unanticipated occurrences during data organization, or errors during data cleaning. Consequently, understanding the intricacies of the entire process before receiving the data is crucial, enabling incorporating these factors into the evaluation.