Solar Farm CSV Walkthrough¶

End-to-end quality control on multi-tag solar SCADA data exported as CSV.

What you will build¶

A short script that:

1. Loads one week of hourly solar farm tags
2. Runs built-in QC rules (or a YAML file)
3. Prints a per-tag quality summary
4. Opens a Plotly quality timeline
5. Writes a self-contained HTML report

Example data shape¶

Use long-format SCADA data — one row per timestamp per tag:

Typical tags for a single-axis tracking plant:

Step 1 — Install and import¶

# Requires timeseries-qc 0.5.0
import tsqc
import pandas as pd

print(tsqc.__version__)  # expect 0.5.0

Step 2 — Load the CSV¶

df = pd.read_csv(
    "solar_farm.csv",
    parse_dates=["timestamp"],
)

print(df.head())
print(df["tag_name"].unique())
print(f"Rows: {len(df)}, date range: {df['timestamp'].min()} → {df['timestamp'].max()}")

CSV exports are almost always timezone-naive. Pass assume_tz in the next step so windows, gaps, and charts stay consistent.

Step 3 — Run the quality check¶

Minimal (built-in defaults)¶

result = tsqc.check(df, assume_tz="UTC")

With plant-specific YAML rules¶

result = tsqc.check(
    df,
    rules="solar_rules.yaml",
    assume_tz="UTC",
)

Example solar_rules.yaml:

default_rules:
  - check: null
    level: bad

tag_rules:
  INVERTER.MW:
    - check: range
      min: 0.0
      max: 11.0
      level: bad
    - check: delta
      max_delta: 5.0
      level: sus
    - check: flatline
      window: 3h
      min_delta: 0.05
      level: sus

  MET.IRRADIANCE:
    - check: range
      min: 0.0
      max: 1050.0
      level: bad
    - check: delta
      max_delta: 400.0
      level: sus
    - check: flatline
      window: 2h
      min_delta: 1.0
      level: sus

  TRACKER.ANGLE:
    - check: range
      min: -90.0
      max: 90.0
      level: bad
    - check: delta
      max_delta: 30.0
      level: sus
    - check: flatline
      window: 4h
      min_delta: 1.0
      level: sus

Tag rules add to default_rules; they do not replace them.

Step 4 — Inspect the summary¶

summary = result.summary()
print(summary)

summary() returns a DataFrame sorted by pct_bad descending:

critical = summary[summary["pct_bad"] > 5.0]
print(critical[["tag_name", "pct_bad", "n_bad"]])

For contiguous issue runs (start/end/duration/reasons):

print(result.issue_summary())

Step 5 — Plot the quality timeline¶

fig = result.plot(title="Solar Farm QC — Week of 2026-06-01")
fig.show()

Each tag is a horizontal bar colored by good / sus / bad. Hover tooltips show reasons such as flatline @ 42.5000 or null values. Timestamps display in the same timezone you passed via assume_tz (UTC here).

Clip to a daylight window if needed:

fig = result.plot(
    tags=["INVERTER.MW", "MET.IRRADIANCE"],
    start="2026-06-03T06:00:00",
    end="2026-06-03T20:00:00",
    title="Daylight QC slice",
)
fig.show()

Step 6 — Export an HTML report¶

result.export_report(
    "solar_farm_qc_report.html",
    title="Solar Farm QC Report",
)

The report is self-contained (no CDN required) and suitable for emailing ops or attaching to a ticket.

Complete script¶

import tsqc
import pandas as pd

df = pd.read_csv("solar_farm.csv", parse_dates=["timestamp"])

result = tsqc.check(
    df,
    rules="solar_rules.yaml",
    assume_tz="UTC",
)

print(result.summary())
print(result.issue_summary())

# Optional: timestamp gaps / duplicates / drift
print(result.check_timestamps(expected_freq="1h"))

result.plot(title="Solar Farm QC").show()
result.export_report("solar_farm_qc_report.html", title="Solar Farm QC Report")

Common pitfalls¶

Next steps¶

• YAML Rules From Scratch — author all five built-in checks
• OSIsoft PI Export — merge historian quality codes
• Quickstart — five-line overview
• Visualization — plot options in depth