Visualize Time Series Data Python Scatter Plot Trendline

Data visualization infographic titled "Scatter Plot Trendline" using SANKEY FLOW as the dominant visual element to show progress through a pipeline, while integrating a secondary scatter plot with trendline as an inset analytical panel. Main composition: a left-to-right sankey pipeline with 5 sequential stages labeled in English: "Raw Data", "Cleaning", "Feature Engineering", "Modeling", "Evaluation". Use realistic plausible illustrative values flowing through the stages: 1,000 -> 820 -> 680 -> 540 -> 510, with proportional flow widths clearly showing retention and drop-off. Add visible side-loss branches labeled "Removed Outliers" 120, "Missing Values" 60, "Irrelevant Features" 140, "Failed Runs" 30. Include sharp English labels, stage totals, clean node captions, and small percentage annotations. Beneath or beside the sankey, include a compact scatter plot inset titled "Trend Over Time" with x-axis label "Time", y-axis label "Performance Score", full tick marks, non-truncated scale, and a clear upward regression line across about 12 plotted points with illustrative values rising plausibly from 42 to 78 over time. Ensure all axes, ticks, legends, and metric names are sharp and fully readable in English. Add 4 key insight callouts with small icons: 1) headline number "51%" with text "Final usable output retained from the starting volume" and a pipeline icon; 2) headline number "+36" with text "Performance score improves across the time series trendline" and a line-chart icon; 3) headline number "140" with text "Largest reduction happens during feature filtering" and a filter icon; 4) headline number "82%" with text "Most early-stage records survive cleaning" and a checkmark icon. Add a small legend in English: "Main flow", "Stage losses", "Trendline", "Observed points". Add a bottom source/data-note strip in English reading: "Data note: Illustrative example. Figures are illustrative and not from a cited external source." Visual style: retro 1970s magazine charts, editorial blue & cream palette with muted navy, dusty teal, faded cobalt, warm cream background, subtle paper grain, restrained orange accent for losses, vintage print texture, geometric icons, elegant serif title plus clean sans-serif labels. Mood: analytical, confident, progress-focused, nostalgic editorial. Include editorial data journalism illustration, FT / Bloomberg-grade chart aesthetics, vector-clean infographic layout. All text MUST be written in English (array). Every heading, label, caption, legend and metric name in the image must be in English — not English. Spell each English word correctly using English characters and diacritics. Numbers stay as digits, no fake authoritative sources cited, no watermarks Numbers labeled "illustrative" unless the user supplied specific sourced data. No fake authoritative sources cited (do not invent "Source: Reuters 2025" — use "Illustrative example" instead). No misleading axis truncation or scale manipulation.