usefulweb.site Time-Series Explorer

Charts & data

Time-Series Explorer

Resample, smooth and decompose date/value data: trend, R², seasonality, YoY change and anomalies.

Max 20 MB. One date column plus one or more value columns (wide), or date, series and value columns (long).

First row = headers. Dates like 2026-03-01, 01.03.2026, 03/01/2026, 2026-03 or Mar 2026.

Empty: the first date column.

Comma-separated; empty: all numeric columns (max 8).

Optional: a column naming the series of each row; then the first value column holds the values.

points

z-score: flag |z| above this (default 3). IQR: fences at k × IQR (default 1.5).

Try an example: Daily visits: weekday pattern, trend and two anomalies · Two products, monthly: seasonality and year-over-year · Daily data resampled to weekly sums

How it works

Upload or paste dated values (one or several series, wide or long format). The explorer resamples to days, weeks, months, quarters or years, adds moving averages (simple, exponential or centred), fits a linear trend with R², estimates weekday or month-of-year seasonality, computes period and year-over-year change and flags anomalies by z-score or IQR, with charts, tables and a CSV export.

  • Dates are parsed per column (ISO, European and US formats, month names, YYYY-MM, YYYY-Qn); rows without a date or number are skipped and counted. Values with the same timestamp are combined with the chosen method.
  • Resampling groups values into calendar periods in UTC (weeks start on Monday, ISO 8601) and combines them by average, sum, first, last, minimum, maximum or median. Empty periods are left out rather than filled.
  • Moving averages: simple = mean of the last w points; exponential = α·value + (1−α)·previous with α = 2/(w+1); centred = mean of w points around each point (2×w for even windows, as in classical decomposition).
  • Trend: ordinary least squares of value on time in days; R² is the share of variance the straight line explains. The slope is also shown per year (× 365.25).
  • Seasonality (an "STL-lite" classical additive decomposition): trend = centred moving average over one period (7 weekdays, 12 months or 4 quarters); seasonal index = average detrended value per weekday/month/quarter, shifted to sum to zero; strength = 1 − var(remainder) / var(detrended), reported once at least two full periods are available. It is a simplification of STL (no LOESS, fixed seasonal pattern).
  • Anomalies are judged on residuals: the decomposition remainder when seasonality is estimated, otherwise the deviation from a centred moving average. z-score flags |z| above the threshold; IQR flags values outside Q1 − k·IQR … Q3 + k·IQR. Anomalies are statistical outliers, not proof of errors.
  • Year-over-year compares the last point with the point on the same calendar date (± a few days, depending on spacing) one year earlier.

For scripts and AI agents

  • JSON API: POST https://timeseriesexplorer.usefulweb.site/api/v1/calculate (same calculation as this page)
  • Interface description: OpenAPI 3.1 · AI guide: llms.txt
  • MCP server (Streamable HTTP): https://timeseriesexplorer.usefulweb.site/mcp · tool name time_series_explorer
Example request
curl -s -X POST https://timeseriesexplorer.usefulweb.site/api/v1/calculate \
  -H 'content-type: application/json' \
  -d '{…see OpenAPI examples…}'

Related tools

  • CSV/Excel Chart Studio – Turn CSV, TSV or Excel data into line, bar, pie, scatter or histogram charts with SVG/PNG export.
  • Historical FX Chart – Chart exchange rates over time from ECB data: min, max, average, change, volatility and CSV export.
Feedback