Evaluate and interpret data

I need clean, auditable analysis of the sensor batch from last month. Import the raw CSVs, remove rows with missing geotags, apply the standard quality filter we agreed with the QA lead, compute the daily mean and standard deviation per measurement type, and generate a single sheet with a chart of the three key metrics over time for the team meeting Thursday.

Before I hand this to the science lead, make the dataset easy to interpret. Put the key metric’s daily mean and confidence interval at the top, highlight dates with anomalous variance, and add a short note explaining which filters removed data and why. Make the chart zoomable and label the major measurement shifts with probable causes pulled from the log.

I ran the analysis and saw variance spike on June 12 across two sensors; the lab log only shows a maintenance window that morning. I worry the maintenance caused a miscalibration but the technician’s note is sparse and I don’t have access to raw telemetry. What is the most likely cause and the minimal checks to confirm whether the June 12 data must be excluded from our report?

We keep losing two days of analysis time because scripts and manual checks are duplicated across analysts before every report. It delays decisions and frustrates the science lead. Which three changes to our data-cleaning routine, documentation, or handoff would cut that time while keeping the audit trail intact?