4,084of 4,246 rows in scope
Data quality

Cleaning & integrity report

Every row in the raw traffic log was profiled before analysis. Rows that fail integrity checks are excluded from all metrics.

Rows ingested
4,246
Rows analysed
4,084
Duplicates removed
46
Rows with missing values
167
Invalid values
32
Duration outliers
236
Cell completeness
9960.94%
Quality score
91/100

Column profile

Type inference, missingness and integrity status per field

ColumnTypeMissingMissing %InvalidUniqueStatusTreatment
user_ididentifier00.00%04,200cleanNo issues detected
session_ididentifier00.00%04,200cleanNo issues detected
timestampdatetime00.00%04,199cleanNo issues detected
variantcategorical00.00%02cleanNo issues detected
landing_pagecategorical00.00%142warning14 invalid
devicecategorical56131.89%03warning56 missing
traffic_sourcecategorical00.00%06cleanNo issues detected
countrycategorical3070.65%08warning30 missing
session_durationnumeric91214.32%181,925warning91 missing · 18 invalid
pages_viewednumeric1330.62%017warning13 missing
clicksnumeric00.00%024cleanNo issues detected
convertedboolean921.20%02warning9 missing

Cleaning pipeline

Applied in order, before any metric is computed

  1. 1
    Parse & type-cast

    Raw CSV strings are coerced to numbers, dates and booleans; unparseable values are recorded as invalid rather than silently zeroed.

  2. 2
    Deduplicate

    Repeated session identifiers are collapsed, keeping the first occurrence — 46 rows removed.

  3. 3
    Integrity checks

    Rows missing a variant assignment or a conversion flag, and rows with negative durations, are marked invalid and excluded.

  4. 4
    Outlier flagging

    Session durations beyond 1.5×IQR are flagged (236) but kept, so they can be included or excluded from the filter bar.

  5. 5
    Metric computation

    Conversion rates, confidence intervals and hypothesis tests are computed only on the surviving rows.