How would you approach analyzing a dataset with missing values?

In my final year project, I analyzed survey data collected from students about their study habits. During the data cleaning phase, I found that some respondents had skipped questions, resulting in missing values. To address this, I first assessed the extent of missing data and discovered it was around 10%. I decided to use mean imputation for questions where responses were quantitative, as it preserved the dataset size without skewing results too much. For qualitative data, I categorized the responses and used the mode to fill in gaps. This approach allowed me to maintain the integrity of the dataset, and ultimately, my findings on study habits were well-received during my presentation.

While volunteering with a local non-profit, I assisted in analyzing health data from community workshops. We noticed that some participants had not provided their ages. Instead of discarding those entries, I suggested conducting a quick survey to gather this information directly from them. For those who still didn’t respond, I researched demographic data for our community and used that information to estimate missing ages while ensuring transparency in our methodology. This helped us complete our dataset, and our analysis on community health trends became more robust as a result.

In my first job as a data analyst, I was tasked with analyzing sales data for the last quarter. We encountered several missing values in the product sales figures. I collaborated with the sales team to understand the reasons for these gaps, which turned out to be due to system errors. We decided to implement a data imputation strategy using regression analysis based on other product sales. This method not only filled in the missing values but also provided insights into trends we hadn't seen before. By addressing the missing data effectively, we improved our sales forecasting accuracy.