When dealing with outliers, I always start with visualisation - scatter plots, box plots, or histograms. These help spot extreme values quickly. Then, I check the data across multiple variables to see if the outlier is a mistake, a true anomaly, or just part of natural variation. Does it make sense? If it's a data entry error, I correct or remove it. If it's real but skews the analysis, I might transform the data (e.g., log or square root) to reduce its impact. If it holds important information, I keep it but choose a robust statistical method like median-based analysis to ensure accurate results.

When I come across outliers, I usually start by checking for them using Z-scores or the IQR method, and I like to visualize the data with box plots or histograms to spot anything unusual. If an outlier is just a data entry mistake, I fix or remove it. But if it’s a real value that just happens to be extreme, I think about whether it’s skewing the results. In that case, I might transform the data (like using a log or square root) to reduce the impact. If the outlier holds important information, I leave it but use median-based methods like MAD to make the analysis more reliable. At the end of the day, context matters and it is important to handle each case carefully, considering possible reasons behind the outlier.

5 Number Summary (minimum ,Q1, median, Q3, Maximum) works well with Quantitative data for Identifying the Outliers. Finding the 4 Quartiles Q1, Q2, Q3, Q4. finding IQR (Inter Quartile Range) = Q3-Q1 Now Calculate the Upper Range and Lower Range. Lower Range = Q1 - 1.5*IQR Upper Range = Q3 + 1.5*IQR If you want to check this for any Normal Distribution, Using Empirical rule we know that 99.7% data lies within 3-Standard_Deviations about the Mean. Data present outside this range can be treated as Outliers. Before removing any data points, I try to understand whether the outliers represent genuine variability (e.g., a niche customer segment) or data errors (e.g., incorrect entries).